Pressure self-adaptive adjusting method and system for double bypass valves of hydraulic cooler

Abstract

The invention provides a hydraulic cooler double-bypass-valve pressure self-adaptive adjusting method and system, and relates to the technical field of hydraulic systems.The method comprises the steps that a fluctuation trend characteristic value is calculated by reading pressure data, characteristic matching is carried out to generate an adjusting instruction, temperature and flow data are collected in real time in a monitoring period, and oil viscosity and Reynolds number are calculated; and when a flow state transition control mode is entered, calculating an optimal flow velocity based on a pressure gradient, converting the optimal flow velocity into a double-bypass valve target opening degree combination, and establishing a mapping relation between an opening degree value and a pressure response by utilizing a steady-state working condition database. Accurate pressure control of the hydraulic system is achieved, and the stability of the system is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field, and particularly relates to a hydraulic cooler double-bypass valve pressure self-adaptive regulation method and system. BACKGROUND

[0002] The hydraulic cooler is an important temperature control device in industrial equipment and hydraulic systems, and its main function is to reduce the temperature of hydraulic oil to ensure the normal operation of the hydraulic system. With the improvement of industrial automation, the working environment of the hydraulic system is becoming more and more complex, and higher requirements are put forward for the pressure control and flow regulation of the hydraulic cooler. The traditional hydraulic cooling system usually adopts a single bypass valve or a manual adjustment method, which cannot adapt to the complex and variable working condition requirements.

[0003] Modern hydraulic cooler systems begin to adopt a double-bypass valve structure, which realizes precise control of the hydraulic system pressure and flow through the cooperative work of the two bypass valves. However, the current double-bypass valve control system still has some technical defects and deficiencies: Most of the existing double-bypass valve control systems adopt a fixed parameter control strategy, lack real-time analysis and prediction ability of the system pressure fluctuation trend, and lead to response lag when the working condition changes, which cannot adjust the valve opening degree in time to adapt to the system pressure change.

[0004] The traditional control method does not fully consider the influence of viscosity change caused by hydraulic oil temperature change on flow state characteristics, especially in the flow state transition region, the control precision is significantly reduced, which easily causes system pressure fluctuation or even pressure impact, and affects the service life of the equipment.

[0005] The existing system lacks a self-learning and data accumulation mechanism, and cannot optimize the control parameters according to the historical operation data. Each time when facing similar working conditions, a complex parameter adjustment process needs to be carried out again, which not only is low in efficiency, but also may cause system instability due to improper parameter setting, thereby increasing the complexity of system maintenance and operation. SUMMARY

[0006] The hydraulic cooler double-bypass valve pressure self-adaptive regulation method and system provided by the embodiments of the present application can solve the problems in the prior art.

[0007] In a first aspect, the hydraulic cooler double-bypass valve pressure self-adaptive regulation method is provided, which comprises: reading pressure data from a data buffer of the hydraulic cooler, calculating a current pressure fluctuation trend characteristic value, and performing characteristic matching between the current pressure fluctuation trend characteristic value and a pressure fluctuation characteristic value in historical data to generate a pressure regulation initial instruction; and sending opening degree regulation signals to a first bypass valve and a second bypass valve according to the pressure regulation initial instruction, and starting a pressure response monitoring timer at the same time. In the monitoring period of the pressure response monitoring timer, real-time acquisition of temperature sensor data and flow sensor data in the hydraulic cooler, calculation of the current hydraulic oil viscosity value according to the temperature sensor data, and calculation of the Reynolds number of each pipeline section in combination with the flow sensor data; when the Reynolds number of any pipeline section exceeds the flow state transition threshold value, entering the flow state transition control mode, calculating the optimal flow rate based on the pressure gradient of the upstream and downstream of the pipeline section, and converting the optimal flow rate into the target opening combination of the first bypass valve and the second bypass valve; Sending an opening dynamic adjustment compensation signal to the first bypass valve and the second bypass valve to gradually adjust the opening thereof to the target opening combination; in the adjustment process, calculating the pressure fluctuation rate in real time according to the opening error change rate, and executing the network output gain coefficient; when the pressure fluctuation rate is lower than the steady-state fluctuation threshold value for a continuous period, storing the current first bypass valve opening value and the current second bypass valve opening value in the steady-state working condition database; taking the opening value in the steady-state working condition database as the reference, establishing a mapping relationship between the opening value and the pressure response, which is used for opening prediction in the next round of pressure adjustment.

[0008] Reading the pressure data from the data buffer of the hydraulic cooler, calculating the current pressure fluctuation trend characteristic value, and performing feature matching between the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic value in the historical data to generate a pressure adjustment initial instruction, including: Obtaining the pressure data of the hydraulic cooler, performing weighted summation on the pressure standard deviation, pressure skewness coefficient and pressure kurtosis coefficient of the pressure data as the current pressure fluctuation trend characteristic value; Calculating the Mahalanobis distance of the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic value in the historical data, multiplying the difference vector of the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic value in the historical data by the inverse matrix of the feature covariance matrix, and the Mahalanobis distance is the square root of the product; Selecting the historical adjustment instruction corresponding to the pressure fluctuation characteristic value in the historical data with the smallest Mahalanobis distance, multiplying the historical adjustment instruction by a correction coefficient to obtain a current adjustment instruction, wherein the correction coefficient is composed of the product of the Mahalanobis distance divided by a distance reference value and a compensation factor, and the sum of 1; Performing the current adjustment instruction to adjust the pressure of the hydraulic cooler.

[0009] When the Reynolds number of any pipeline section exceeds the flow state transition threshold value, entering the flow state transition control mode, calculating the optimal flow rate based on the pressure gradient of the upstream and downstream of the pipeline section, and converting the optimal flow rate into the target opening combination of the first bypass valve and the second bypass valve, including: Based on the flow rate, the pipe diameter, the hydraulic oil density and the dynamic viscosity at the reference temperature of the hydraulic cooler pipeline, a temperature correction coefficient is calculated according to the difference between the current temperature and the reference temperature, the dynamic viscosity at the current temperature is obtained by using the temperature correction coefficient, and the product of the hydraulic oil density, the flow rate and the pipe diameter is divided by the dynamic viscosity at the current temperature to obtain the Reynolds number of the pipeline; A local pressure gradient of the pipeline is obtained, a first ratio is obtained by dividing the Reynolds number by a critical Reynolds number, a second ratio is obtained by dividing the absolute value of the local pressure gradient by a critical pressure gradient, and a weighted sum of the first ratio and the second ratio is taken as a flow regime transition criterion; When the flow regime transition criterion is greater than 1, a target flow rate at which the energy loss is the smallest is calculated based on the friction loss coefficient, the local loss coefficient and the pipe segment length of the pipeline; An optimization objective function of the first bypass valve opening degree and the second bypass valve opening degree is constructed according to the target flow rate, the flow coefficient, the hydraulic oil density and the through-flow area, wherein the optimization objective function includes a flow deviation term, an opening degree difference term and an opening degree change rate term; and the first bypass valve target opening degree and the second bypass valve target opening degree are obtained by solving the optimization objective function; A time constant is dynamically calculated based on the deviation value of the Reynolds number and the critical Reynolds number, and the first bypass valve target opening degree and the second bypass valve target opening degree are gradually adjusted according to the time constant to realize smooth transition of the flow regime.

[0010] The current first bypass valve opening degree value and the current second bypass valve opening degree value are stored in a steady state working condition database; and a mapping relationship between the opening degree value and the pressure response is established based on the opening degree values in the steady state working condition database, which is used for opening degree pre-judgment in the next round of pressure regulation, including: A plurality of groups of training samples of the first bypass valve opening degree, the second bypass valve opening degree and the system pressure in the steady state working condition database are obtained; A nonlinear mapping model of the opening degree and the pressure is established based on the training samples; a base function center and an expansion parameter are determined, and a radial basis function value of each group of training samples is calculated according to the first bypass valve opening degree and the second bypass valve opening degree; wherein the radial basis function value is obtained by dividing the sum of squares of the differences between the first bypass valve opening degree and the second bypass valve opening degree and the base function center by the square of the expansion parameter, and taking a negative exponent of the quotient value; The product of the radial basis function value and a weight coefficient is accumulated to obtain a nonlinear mapping function; the sum of squares of the difference between the output value of the nonlinear mapping function and the system pressure is calculated, the sum of squares of the weight coefficient is calculated, and a weighted sum of the sum of squares and the sum of squares of the weight coefficient is taken as a loss function; obtaining a target pressure value, taking the difference between the target pressure value and the output value of the nonlinear mapping function as a target function; calculating the partial derivative of the target function with respect to the first bypass valve opening and the partial derivative of the target function with respect to the second bypass valve opening; According to the partial derivative of the first bypass valve opening and the partial derivative of the second bypass valve opening, the first bypass valve opening and the second bypass valve opening are iteratively updated to obtain the first bypass valve target opening and the second bypass valve target opening.

[0011] Sending opening dynamic adjustment compensation signals to the first bypass valve and the second bypass valve to gradually adjust their openings to the target opening combination; during the adjustment process, the pressure fluctuation rate is calculated in real time according to the opening error change rate, and the network output gain coefficient is executed, including: Obtaining a system pressure signal sequence, a first bypass valve opening sequence and a second bypass valve opening sequence, and constructing the system pressure signal sequence, the first bypass valve opening sequence and the second bypass valve opening sequence into a historical data sequence; Calculating the system pressure fluctuation rate, obtaining the difference between the first bypass valve target opening and the first bypass valve current opening to obtain the first opening error, obtaining the difference between the second bypass valve target opening and the second bypass valve current opening to obtain the second opening error, and calculating the change rate of the first opening error and the change rate of the second opening error; The system pressure fluctuation rate, the first opening error, the second opening error, the change rate of the first opening error, the change rate of the second opening error and the historical data sequence are constructed into a state vector; The state vector is input into an execution network, and the proportional gain coefficient, the differential gain coefficient and the fluctuation suppression coefficient are output through the execution network; wherein the proportional gain coefficient and the system pressure fluctuation rate are negatively exponentially related, the differential gain coefficient and the system pressure fluctuation rate are negatively exponentially related, and the fluctuation suppression coefficient and the system pressure fluctuation rate are positively correlated.

[0012] Sending opening dynamic adjustment compensation signals to the first bypass valve and the second bypass valve to gradually adjust their openings to the target opening combination, including: According to the proportional gain coefficient, the differential gain coefficient and the fluctuation suppression coefficient, first and second compensation signals are generated; the state vector and the first compensation signal are input into an evaluation network to obtain a first state value, and the state vector and the second compensation signal are input into the evaluation network to obtain a second state value; According to the first state value and the second state value, a reward value is calculated, and the state vector, the first compensation signal, the second compensation signal, the reward value and the next time state vector are stored in an experience pool; The sample is obtained from the experience pool to optimize the training of the execution network and the evaluation network; the first compensation signal and the second compensation signal are smoothed to obtain a smoothed compensation signal; and the opening degree of the first bypass valve and the opening degree of the second bypass valve are adjusted according to the smoothed compensation signal.

[0013] In a second aspect of the embodiment of the present application, a hydraulic cooler double-bypass valve pressure self-adaptive adjustment system is provided, comprising: A first unit is configured to read pressure data from a data buffer of the hydraulic cooler, calculate a current pressure fluctuation trend characteristic value, and perform characteristic matching between the current pressure fluctuation trend characteristic value and pressure fluctuation characteristic values in historical data to generate a pressure adjustment initial instruction; and send opening degree adjustment signals to the first bypass valve and the second bypass valve according to the pressure adjustment initial instruction, and simultaneously start a pressure response monitoring timer. A second unit is configured to collect temperature sensor data and flow sensor data in the hydraulic cooler in real time within a monitoring period of the pressure response monitoring timer, calculate a current hydraulic oil viscosity value according to the temperature sensor data, and calculate Reynolds numbers of each pipeline section in combination with the flow sensor data; when the Reynolds number of any pipeline section exceeds a flow state transition threshold value, enter a flow state transition control mode, calculate an optimal flow rate based on pressure gradients of the upstream and downstream of the pipeline section, and convert the optimal flow rate into a target opening degree combination of the first bypass valve and the second bypass valve. A third unit is configured to send opening degree dynamic adjustment compensation signals to the first bypass valve and the second bypass valve, so that the opening degrees thereof are gradually adjusted to the target opening degree combination; in the adjustment process, calculate a pressure fluctuation rate in real time according to an opening degree error change rate, and output a gain coefficient through an execution network; when the pressure fluctuation rate is lower than a steady-state fluctuation threshold value for a continuous period, store a current first bypass valve opening degree value and a current second bypass valve opening degree value in a steady-state working condition database; and establish a mapping relationship between the opening degree values and pressure responses based on the opening degree values in the steady-state working condition database, for opening degree prediction in the next round of pressure adjustment.

[0014] In a third aspect of the embodiment of the present application, An electronic device is provided, comprising: a processor; a memory for storing processor-executable instructions; The processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0015] In a fourth aspect of the embodiment of the present application, A computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0016] The beneficial effects of the present application are as follows: The present application realizes accurate control of the hydraulic system pressure, improves the stability and reliability of the system through the double bypass valve pressure self-adaptive adjustment method.

[0017] The present application adopts the real-time calculation of the hydraulic oil viscosity and Reynolds number, can accurately identify the flow state transition, timely enters the flow state transition control mode, calculates the optimal flow rate according to the pressure gradient and converts it into a valve opening degree combination, effectively prevents the pressure fluctuation and instability phenomenon of the hydraulic system under different working conditions.

[0018] The present application establishes the mapping relationship database of the opening value and pressure response, forms a closed-loop self-learning mechanism through the historical data feature matching and steady-state working condition record, makes the system be able to predict the opening value according to the historical experience, significantly shortens the pressure regulation time, reduces the energy consumption, and prolongs the service life of the hydraulic system. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 It is a flowchart of the double bypass valve pressure self-adaptive adjustment method of the hydraulic cooler of the embodiment of the present application. Figure 2 It is a system architecture schematic diagram of the hydraulic cooler flow state transition control system. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the embodiment of the present application more clear, the technical scheme in the embodiment of the present application will be described clearly and completely in combination with the drawings in the embodiment of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] The technical scheme of the present application will be described in detail in the following specific embodiments. The following several specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in some embodiments.

[0022] Figure 1 It is a flowchart of the double bypass valve pressure self-adaptive adjustment method of the hydraulic cooler of the embodiment of the present application, as shown in Figure 1 The method comprises: read pressure data from a data buffer of the hydraulic cooler, calculate a current pressure fluctuation trend characteristic value, and perform characteristic matching of the current pressure fluctuation trend characteristic value with pressure fluctuation characteristic values in historical data to generate a pressure regulation initial instruction; according to the pressure regulation initial instruction, send opening degree regulation signals to the first bypass valve and the second bypass valve respectively, and simultaneously start a pressure response monitoring timer; within a monitoring period of the pressure response monitoring timer, collect temperature sensor data and flow sensor data in the hydraulic cooler in real time, calculate a current hydraulic oil viscosity value according to the temperature sensor data, and calculate Reynolds numbers of each pipe section in combination with the flow sensor data; when the Reynolds number of any pipe section exceeds a flow state transition threshold value, enter a flow state transition control mode, calculate an optimal flow rate based on pressure gradients of the upstream and downstream of the pipe section, and convert the optimal flow rate into a target opening degree combination of the first bypass valve and the second bypass valve; send opening degree dynamic regulation compensation signals to the first bypass valve and the second bypass valve to gradually regulate the opening degrees thereof to the target opening degree combination; in the regulation process, calculate a pressure fluctuation rate in real time according to an opening degree error change rate, and execute a network output gain coefficient; when the pressure fluctuation rate is lower than a steady-state fluctuation threshold value for a continuous period, store a current first bypass valve opening degree value and a current second bypass valve opening degree value in a steady-state working condition database; take the opening degree values in the steady-state working condition database as a reference to establish a mapping relationship between the opening degree values and pressure responses, which is used for opening degree prediction in the next round of pressure regulation.

[0023] In an alternative embodiment, reading pressure data from a data buffer of the hydraulic cooler, calculating a current pressure fluctuation trend characteristic value, and performing characteristic matching of the current pressure fluctuation trend characteristic value with pressure fluctuation characteristic values in historical data to generate a pressure regulation initial instruction, comprises: obtaining pressure data of the hydraulic cooler, performing weighted summation of a pressure standard deviation, a pressure skewness coefficient and a pressure kurtosis coefficient of the pressure data as a current pressure fluctuation trend characteristic value; calculating a Mahalanobis distance of the current pressure fluctuation trend characteristic value and pressure fluctuation characteristic values in historical data, multiplying a difference vector of the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic values in historical data by an inverse matrix of a characteristic covariance matrix, and the Mahalanobis distance is a square root of the product; selecting a historical regulation instruction corresponding to a pressure fluctuation characteristic value in historical data with the smallest Mahalanobis distance, multiplying the historical regulation instruction by a correction coefficient to obtain a current regulation instruction, wherein the correction coefficient is composed of a product of the Mahalanobis distance divided by a distance reference value and a compensation factor, and 1; executing the current regulation instruction to regulate the pressure of the hydraulic cooler.

[0024] In one embodiment of the present invention, a method for regulating the pressure of a hydraulic cooler is provided. The method reads pressure data from a data buffer of the hydraulic cooler, calculates the current pressure fluctuation trend characteristic value, and performs feature matching with the pressure fluctuation characteristic value in historical data to generate an initial pressure regulation instruction.

[0025] Pressure data is read from the hydraulic cooler's data buffer. This data contains pressure sampling points over a period of time (e.g., 10 minutes). For example, if the system collects 5 pressure data points per second, there are 3,000 pressure data points in 10 minutes. This data is temporarily stored in the data buffer for subsequent analysis. Three statistical characteristics are calculated for the read pressure data: the pressure standard deviation, the pressure skewness coefficient, and the pressure kurtosis coefficient. The pressure standard deviation reflects the degree of data dispersion. For example, for the 3,000 data points read, the calculated standard deviation is 0.358 MPa. The pressure skewness coefficient reflects the symmetry of the data distribution and is calculated to be 0.127. The pressure kurtosis coefficient reflects the angularity of the data distribution and is calculated to be 2.874.

[0026] The weighted sum of these three eigenvalues ​​yields the current pressure fluctuation trend eigenvalue. In this example, the standard deviation is weighted 0.5, the skewness coefficient is weighted 0.3, and the kurtosis coefficient is weighted 0.2. Based on these weights, the current pressure fluctuation trend eigenvalue is calculated as 0.358 × 0.5 + 0.127 × 0.3 + 2.874 × 0.2 = 0.753. This eigenvalue represents the comprehensive characteristics of the current hydraulic cooler pressure fluctuation.

[0027] The current pressure fluctuation trend characteristic values ​​are compared with those stored in the historical database, and the similarity is assessed by calculating the Mahalanobis distance. The Mahalanobis distance considers the covariance relationship between each feature and can more accurately reflect the degree of similarity between features. Specifically, multiple sets of historical pressure fluctuation characteristic values ​​are extracted from the historical database, each set containing three indicators: standard deviation, skewness coefficient, and kurtosis coefficient.

[0028] Obtain five sets of historical data from the historical database: the first set is [0.342, 0.119, 2.832], the second set is [0.375, 0.135, 2.913], the third set is [0.320, 0.110, 2.745], the fourth set is [0.390, 0.142, 2.965], and the fifth set is [0.335, 0.122, 2.810]. Calculate the covariance matrix of these historical data sets and find its inverse. The inverse matrix of the characteristic covariance matrix is ​​[[15.24, -8.37, 1.25], [-8.37, 25.63, -2.16], [1.25, -2.16, 8.92]].

[0029] For each group of historical data, a difference vector is calculated between it and the current pressure fluctuation trend eigenvalue. For example, the first group of historical data, the difference vector is [0.358-0.342, 0.127-0.119, 2.874-2.832] = [0.016, 0.008, 0.042]. The difference vector is multiplied by the inverse of the eigen-covariance matrix, and the square root is calculated to obtain the Mahalanobis distance. For the first group of historical data, the calculated Mahalanobis distance is 0.312.

[0030] Similarly, the Mahalanobis distance between the current pressure fluctuation trend eigenvalue and other historical data groups is calculated, which are 0.487, 0.653, 0.712, and 0.295 respectively. Among them, the Mahalanobis distance 0.295 corresponding to the fifth group of historical data is the smallest, indicating that the current pressure fluctuation trend is most similar to the fifth group of historical data.

[0031] According to the calculation results of the Mahalanobis distance, the historical adjustment instruction corresponding to the historical data with the smallest Mahalanobis distance is selected. Assuming that the historical adjustment instruction corresponding to the fifth group of historical data is "increase coolant flow by 0.5 L / min, and reduce pressure valve opening by 15%". In order to adapt to the current working condition, the historical adjustment instruction needs to be modified.

[0032] The calculation of the correction coefficient is based on the Mahalanobis distance and a preset distance reference value. The distance reference value represents an ideal matching degree, which is set to 0.25 in this embodiment. The calculation formula of the correction coefficient is: 1 + Mahalanobis distance / distance reference value × compensation factor. When the compensation factor is -0.2, the correction coefficient is 1 + 0.295 / 0.25 × (-0.2) = 0.764.

[0033] The historical adjustment instruction is multiplied by the correction coefficient to obtain the current adjustment instruction. For the instruction to increase the coolant flow, the modified value is 0.5 × 0.764 = 0.382 L / min; for the instruction to reduce the pressure valve opening, the modified value is 15% × 0.764 = 11.46%. Therefore, the current adjustment instruction is "increase coolant flow by 0.382 L / min, and reduce pressure valve opening by 11.46%".

[0034] Executing this adjustment instruction, the system adjusts the coolant flow and pressure valve opening of the hydraulic cooler through the control system. The system controls the speed of the coolant pump through the servo motor to increase the coolant flow by 0.382 L / min; at the same time, the pressure valve is controlled by the stepper motor to reduce the opening by 11.46%. These adjustments will directly affect the pressure of the hydraulic cooler, making it tend to a stable state.

[0035] After executing the adjustment instruction, the system continues to monitor the pressure data of the hydraulic cooler and stores the new pressure data and the corresponding adjustment instruction in the historical database for future more accurate feature matching and pressure adjustment. In this way, the system can continuously learn and optimize the pressure adjustment strategy to improve the operating efficiency and stability of the hydraulic cooler.

[0036] In an alternative embodiment, when the Reynolds number of any pipe section exceeds the flow regime transition threshold, the flow regime transition control mode is entered, the optimal flow rate is calculated based on the pressure gradient upstream and downstream of the pipe section, and the optimal flow rate is converted into the target opening combination of the first bypass valve and the second bypass valve, including: Based on the flow rate, pipe diameter, hydraulic oil density, and dynamic viscosity at the reference temperature of the hydraulic cooler pipeline, a temperature correction coefficient is calculated according to the difference between the current temperature and the reference temperature, the dynamic viscosity at the current temperature is obtained using the temperature correction coefficient, and the product of the hydraulic oil density, flow rate, and pipe diameter is divided by the dynamic viscosity at the current temperature to obtain the Reynolds number of the pipeline; The local pressure gradient of the pipeline is obtained, the first ratio is obtained by dividing the Reynolds number by the critical Reynolds number, the second ratio is obtained by dividing the absolute value of the local pressure gradient by the critical pressure gradient, and the weighted sum of the first ratio and the second ratio is used as the flow regime transition criterion; When the flow regime transition criterion is greater than 1, the target flow rate at which the energy loss is minimized is calculated based on the friction loss coefficient, local loss coefficient, and pipe section length of the pipeline; According to the target flow rate, flow coefficient, hydraulic oil density, and through-flow area, an optimization objective function of the first bypass valve opening and the second bypass valve opening is constructed, wherein the optimization objective function includes a flow deviation term, an opening difference term, and an opening change rate term; the first bypass valve target opening and the second bypass valve target opening are obtained by solving the optimization objective function; The time constant is dynamically calculated based on the deviation value of the Reynolds number and the critical Reynolds number, and the first bypass valve target opening and the second bypass valve target opening are gradually adjusted according to the time constant to achieve smooth transition of the flow regime.

[0037] Figure 2 The hydraulic cooler flow regime transition control system architecture diagram is shown in the embodiment. When the Reynolds number of any pipe section exceeds the flow regime transition threshold, the system enters the flow regime transition control mode, which can calculate the optimal flow rate based on the pressure gradient upstream and downstream of the pipe section, and convert the optimal flow rate into the target opening combination of the first bypass valve and the second bypass valve.

[0038] To determine the Reynolds number of the pipe segment, the system calculates based on the flow rate, pipe diameter, hydraulic oil density, and dynamic viscosity at reference temperature of the hydraulic cooler pipe. Assume the flow rate of the pipe is 2.5 m / s, the pipe diameter is 0.02 m, the hydraulic oil density is 850 kg / m3, and the dynamic viscosity at reference temperature of 25 °C is 0.04 Pa-s. When the actual temperature is 40 °C, the system calculates the temperature correction coefficient to be 0.85. Apply the correction coefficient to the dynamic viscosity at reference temperature to obtain the dynamic viscosity at current temperature, which is 0.034 Pa-s. Divide the product of the hydraulic oil density 850 kg / m3, flow rate 2.5 m / s, and pipe diameter 0.02 m by the dynamic viscosity at current temperature 0.034 Pa-s to obtain the Reynolds number of the pipe, which is 1250.

[0039] The system then obtains the local pressure gradient of the pipe, for example, the measured pressure gradient is 15000 Pa / m. The system divides the Reynolds number 1250 by the critical Reynolds number 1000 to obtain a first ratio 1.25, and divides the absolute value of the local pressure gradient 15000 Pa / m by the critical pressure gradient 12000 Pa / m to obtain a second ratio 1.25. The system calculates the weighted sum of the first ratio and the second ratio to be 1.25 using the weighted coefficients 0.6 and 0.4, respectively, which is used as the flow regime transition criterion.

[0040] Since the flow regime transition criterion 1.25 is greater than 1, the system confirms that it enters the flow regime transition control mode. At this time, the system calculates the target flow rate at which the energy loss is the smallest based on the frictional loss coefficient 0.03, local loss coefficient 1.5, and pipe segment length 2 m of the pipe. By considering the frictional loss and local loss comprehensively, the system calculates the target flow rate to be 2.2 m / s.

[0041] According to the target flow rate 2.2 m / s, flow coefficient 0.75, hydraulic oil density 850 kg / m3, and through-flow area 0.0003 m2, the system constructs an optimization objective function of the first bypass valve opening and the second bypass valve opening. The objective function includes a flow deviation term, an opening difference term, and an opening change rate term. The flow deviation term is used to ensure that the difference between the actual flow rate and the target flow rate is the smallest; the opening difference term is used to balance the openings of the two bypass valves to avoid excessive load on a single valve; and the opening change rate term is used to limit the rapid change of the valve opening to prevent system oscillation.

[0042] The system assigns weight coefficients 0.5, 0.3, and 0.2 to the three terms, respectively, and solves the optimization objective function. In the current state, the current opening of the first bypass valve is 40%, and the current opening of the second bypass valve is 35%. Through numerical optimization method, the system calculates the target opening of the first bypass valve to be 45%, and the target opening of the second bypass valve to be 42%.

[0043] In view of the need for smooth transition of flow regime, the system dynamically calculates the time constant based on the deviation of Reynolds number from the critical Reynolds number. The deviation of the current Reynolds number 1250 from the critical Reynolds number 1000 is 250, and the system calculates the time constant as 2 seconds based on the deviation. According to the time constant, the system gradually adjusts the target opening of the first bypass valve and the target opening of the second bypass valve.

[0044] Specifically, the system updates the actual opening of the valve every 100 milliseconds. In the first step, the opening of the first bypass valve is adjusted from 40% to 41%, and the opening of the second bypass valve is adjusted from 35% to 36%. In the second step, the opening of the first bypass valve is adjusted from 41% to 42%, and the opening of the second bypass valve is adjusted from 36% to 37%. This process is repeated until the target opening is reached. This gradual adjustment method ensures smooth transition of flow regime and avoids system oscillation and impact.

[0045] During the entire flow regime transition control process, the system continuously monitors the Reynolds number and pressure gradient of the pipeline. When the flow regime transition criterion falls below 0.9 (less than 1), the system exits the flow regime transition control mode and returns to normal control mode. The system also has a failsafe mechanism. When an abnormal situation is detected, such as a sudden pressure rise exceeding the safety threshold of 20000 Pascal / meter, the system immediately closes both bypass valves to a safe position (e.g., 20% opening) and triggers an alarm.

[0046] The above-mentioned flow regime transition control mode achieves smooth control of the hydraulic cooling system during flow regime change by accurately calculating the optimal flow rate and converting it into the appropriate valve opening combination, effectively reducing energy loss and improving system efficiency, while avoiding system instability and safety hazards caused by sudden flow regime changes.

[0047] In an alternative embodiment, the current first bypass valve opening value and the current second bypass valve opening value are stored in a steady-state operating condition database. Based on the opening values in the steady-state operating condition database, a mapping relationship between opening values and pressure responses is established for opening prediction during the next round of pressure regulation, including: Obtain a plurality of training samples of first bypass valve opening, second bypass valve opening, and system pressure in the steady-state operating condition database; Based on the training samples, a nonlinear mapping model of opening and pressure is established. The center of the base function and the expansion parameter are determined, and the radial basis function value of each training sample is calculated based on the first bypass valve opening and the second bypass valve opening. The radial basis function value is obtained by dividing the sum of the squares of the differences between the first bypass valve opening and the second bypass valve opening and the center of the base function by the square of the expansion parameter, and taking the negative exponent of the quotient value. The product of the radial basis function value and the weight coefficient is accumulated to obtain a nonlinear mapping function; the sum of squares of the difference between the output value of the nonlinear mapping function and the system pressure is calculated, the sum of squares of the weight coefficients is calculated, and the weighted sum of the sum of squares of the difference and the sum of squares of the weight coefficients is taken as a loss function; A target pressure value is obtained, and the difference between the target pressure value and the output value of the nonlinear mapping function is taken as an objective function; the partial derivative of the objective function with respect to the first bypass valve opening and the partial derivative of the objective function with respect to the second bypass valve opening are calculated; The first bypass valve opening and the second bypass valve opening are iteratively updated according to the partial derivative of the first bypass valve opening and the partial derivative of the second bypass valve opening, to obtain a first bypass valve target opening and a second bypass valve target opening.

[0048] In this embodiment, in order to realize accurate control of the system pressure, a nonlinear mapping method based on radial basis function is proposed to predict the relationship between the bypass valve opening and the system pressure.

[0049] When the system reaches a steady state operating condition, the controller stores the current first bypass valve opening value and the current second bypass valve opening value in the steady state operating condition database. These opening values, together with the corresponding system pressure values, constitute training samples for establishing the mapping relationship between the opening values and the pressure response. The system obtains multiple groups of training samples from the steady state operating condition database, each group of samples containing the first bypass valve opening, the second bypass valve opening and the corresponding system pressure value. For example, the database stores multiple groups of data such as {first bypass valve opening = 25%, second bypass valve opening = 35%, system pressure = 5.2 MPa}, {first bypass valve opening = 30%, second bypass valve opening = 40%, system pressure = 4.8 MPa}.

[0050] Based on these training samples, the system establishes a nonlinear mapping model of opening and pressure. The model adopts a radial basis function network structure, and first needs to determine the basis function center and the expansion parameter. The basis function center can be determined by selecting representative sample points from the training samples, for example, 10 sample points with relatively uniform opening values can be selected as the basis function center; the expansion parameter determines the width of the radial basis function, and can be set to 0.8 times the average distance between adjacent center points. When the average distance between center points is 5 percentage points, the expansion parameter can be set to 4.

[0051] For each training sample, the system calculates the radial basis function value according to its first and second bypass valve openings. The calculation process is: square the difference between the sample's first and second bypass valve openings and each basis function center, sum them up, divide by the square of the spread parameter, and take the negative exponent of the result. For a sample {first bypass valve opening = 25%, second bypass valve opening = 35%}, if the basis function center is {20%, 30%} and the spread parameter is 4, the squared difference sum is (25-20)²+(35-30)²=50, divided by the square of the spread parameter 16 to get 3.125, and the negative exponent to get a radial basis function value of about 0.044.

[0052] The system assigns each basis function a weight coefficient, which can be initially set to a random small number, such as a random number between 0.1 and 0.5. By multiplying the output value of each basis function by the corresponding weight coefficient and accumulating them, the nonlinear mapping function is obtained. The radial basis function values corresponding to the 10 basis functions are [0.044, 0.023, 0.067,...], and the weight coefficients are [0.3, 0.4, 0.2,...], and the output value of the mapping function is 0.044×0.3+0.023×0.4+0.067×0.2+...

[0053] To optimize this mapping model, the system defines a loss function, which consists of two parts: the first part is the squared sum of the difference between the output value of the nonlinear mapping function and the actual system pressure; the second part is the square sum of the weight coefficients multiplied by a regularization coefficient (0.01). By minimizing this loss function, the optimal weight coefficients can be found. If the output value of the mapping function for a certain sample is 5.3 MPa, and the actual system pressure is 5.2 MPa, the squared difference is 0.01; accumulate the squared difference of all samples, and add the square sum of the weight coefficients multiplied by the regularization coefficient to get the total loss value.

[0054] The system uses gradient descent method to optimize the weight coefficients, iteratively updating until the loss function converges or reaches the preset number of iterations (1000 times). The optimized model can accurately reflect the relationship between the bypass valve opening and the system pressure.

[0055] When the system pressure needs to be adjusted, the controller obtains the target pressure value, such as adjusting the system pressure to 5.0 MPa. The controller takes the difference between the target pressure value and the output value of the nonlinear mapping function as the objective function. In order to find the bypass valve opening that can achieve the target pressure, the controller calculates the partial derivative of the objective function with respect to the first and second bypass valve openings.

[0056] These partial derivatives indicate the direction and magnitude of the pressure change caused by the opening change. The partial derivative of the first bypass valve opening is -0.05, indicating that increasing the opening by 1 percentage point will reduce the pressure by about 0.05 MPa; the partial derivative of the second bypass valve opening is -0.03, indicating that increasing the opening by 1 percentage point will reduce the pressure by about 0.03 MPa.

[0057] Based on these partial derivatives, the controller uses the gradient descent method to update the bypass valve opening. The current opening is {first bypass valve = 28%, second bypass valve = 38%}, and the learning rate is set to 0.5, so the opening after the first iteration is updated to {first bypass valve = 28 + 0.5 × (-0.05) = 27.975%, second bypass valve = 38 + 0.5 × (-0.03) = 37.985%}. The controller continues to iterate and update the opening value until the convergence condition is met or the maximum number of iterations (50) is reached, and finally obtains the first bypass valve target opening and the second bypass valve target opening that can achieve the target pressure.

[0058] Through this method, the system can quickly and accurately predict the required valve opening under different working conditions, achieve precise pressure control, and reduce fluctuations and stabilization time during the adjustment process.

[0059] In an alternative embodiment, a dynamic opening adjustment compensation signal is sent to the first bypass valve and the second bypass valve, so that the opening of the first bypass valve and the second bypass valve is gradually adjusted to the target opening combination; during the adjustment process, the pressure fluctuation rate is calculated in real time according to the opening error change rate, and the network output gain coefficient is executed, including: Obtain a system pressure signal sequence, a first bypass valve opening sequence and a second bypass valve opening sequence, and construct the system pressure signal sequence, the first bypass valve opening sequence and the second bypass valve opening sequence into a historical data sequence; Calculate the system pressure fluctuation rate, obtain the difference between the first bypass valve target opening and the first bypass valve current opening to get the first opening error, obtain the difference between the second bypass valve target opening and the second bypass valve current opening to get the second opening error, and calculate the change rate of the first opening error and the change rate of the second opening error; Construct the system pressure fluctuation rate, the first opening error, the second opening error, the change rate of the first opening error, the change rate of the second opening error and the historical data sequence into a state vector; Input the state vector into the execution network, and output the proportional gain coefficient, the differential gain coefficient and the fluctuation suppression coefficient through the execution network; wherein the proportional gain coefficient and the system pressure fluctuation rate are in negative exponential relationship, the differential gain coefficient and the system pressure fluctuation rate are in negative exponential relationship, and the fluctuation suppression coefficient and the system pressure fluctuation rate are in positive correlation.

[0060] In this embodiment, the opening degree dynamic adjustment compensation signal is sent to the first bypass valve and the second bypass valve, so that the opening degrees are gradually adjusted to the target opening degree combination, and the pressure fluctuation rate is calculated in real time according to the opening degree error change rate during the adjustment process, and the network output gain coefficient is executed to realize smooth control.

[0061] The control system first acquires a system pressure signal sequence, a first bypass valve opening degree sequence and a second bypass valve opening degree sequence. The system pressure signal sequence is collected by a pressure sensor arranged in the pipeline system, with a sampling frequency of 100 Hz, 100 continuous pressure data points are collected each time to form a pressure signal sequence; the opening degree signals of the first bypass valve and the second bypass valve are collected in real time by a valve position sensor, also with a sampling frequency of 100 Hz, 100 opening degree data points are collected each time to form an opening degree sequence. The three data sequences constitute a historical data sequence, which is used for subsequent real-time calculation and analysis.

[0062] When the system calculates the pressure fluctuation rate, the standard deviation method is used to process the pressure signal sequence. Specifically, the average value of the 100 recently collected pressure data points is calculated, then the deviation square sum of each data point from the average value is calculated, divided by the number of data points, and finally the standard deviation value is obtained by taking the square root. If the average value of the 100 pressure data points is 5.2 MPa, the calculated standard deviation is 0.08 MPa, and the pressure fluctuation rate is 0.08 / 5.2=1.54%.

[0063] The system obtains the opening degree error by comparing the target opening degree with the current opening degree. The target opening degree of the first bypass valve is 85%, and the current opening degree is 72%, so the first opening degree error is 13%; the target opening degree of the second bypass valve is 65%, and the current opening degree is 70%, so the second opening degree error is -5%. The change rate of the opening degree error is calculated by dividing the difference between the opening degree errors of two consecutive samplings by the sampling time interval. If the last sampling time the first bypass valve opening degree error is 15%, and the current is 13%, the sampling interval is 0.01 seconds, then the first opening degree error change rate is (13%-15%) / 0.01=-200% / s; similarly, if the last sampling time the second bypass valve opening degree error is -4%, and the current is -5%, then the second opening degree error change rate is (-5%-(-4%)) / 0.01=-100% / s.

[0064] The system pressure fluctuation rate (1.54%), the first opening degree error (13%), the second opening degree error (-5%), the first opening degree error change rate (-200% / s), and the second opening degree error change rate (-100% / s) are combined with the historical data sequence to form a state vector. The historical data sequence contains pressure and opening degree data in the past 10 seconds and is stored in matrix form. The state vector constructed in this way reflects the current state and recent trend of the system.

[0065] The state vector is input into a pre-trained execution network. The execution network adopts a deep neural network structure, including an input layer, three hidden layers, and an output layer. The number of nodes in the input layer is the same as the dimension of the state vector, the hidden layers contain 64, 32, and 16 neurons respectively, and the output layer has three nodes corresponding to the proportional gain coefficient, the differential gain coefficient, and the fluctuation suppression coefficient. The network is trained by a large amount of historical operation data and can output the optimal control parameters according to the current system state.

[0066] The proportional gain coefficient output by the execution network has a negative exponential relationship with the system pressure fluctuation rate, which means that the larger the pressure fluctuation rate, the smaller the proportional gain coefficient. For example, when the pressure fluctuation rate is 1.54%, the proportional gain coefficient output by the network is 0.65; when the pressure fluctuation rate rises to 3%, the proportional gain coefficient will decrease to 0.25. The differential gain coefficient also has a negative exponential relationship with the system pressure fluctuation rate, for example, when the pressure fluctuation rate is 1.54%, the differential gain coefficient is 0.12; when the pressure fluctuation rate rises to 3%, the differential gain coefficient will decrease to 0.05. The fluctuation suppression coefficient has a positive correlation with the system pressure fluctuation rate, for example, when the pressure fluctuation rate is 1.54%, the fluctuation suppression coefficient is 0.35; when the pressure fluctuation rate rises to 3%, the fluctuation suppression coefficient will increase to 0.7.

[0067] After obtaining the three coefficients, the system generates a dynamic opening degree adjustment compensation signal. The calculation of the compensation signal takes into account the opening degree error, the opening degree error rate, and the three coefficients. For the first bypass valve, its compensation signal is the proportional gain coefficient (0.65) multiplied by the opening degree error (13%) plus the differential gain coefficient (0.12) multiplied by the opening degree error rate (-200% / s), multiplied by (1-fluctuation suppression coefficient (0.35)), resulting in a compensation signal value of 5.02%. The compensation signal calculation method for the second bypass valve is the same.

[0068] The control system sends the calculated compensation signal to the corresponding bypass valve actuator, so that the valve opening degree is adjusted according to the amplitude indicated by the compensation signal. The current opening degree of the first bypass valve is 72%, and the compensation signal is 5.02%, so the valve opening degree is adjusted to 77.02%; the current opening degree of the second bypass valve is 70%, and the compensation signal is assumed to be -3.15%, so the valve opening degree is adjusted to 66.85%. The system repeats the above calculation and adjustment process at a frequency of 10 Hz, so that the valve opening degree gradually approaches the target opening degree while maintaining the stability of the system pressure.

[0069] Through this dynamic adjustment method, the system can adaptively control the adjustment speed and amplitude during valve adjustment, effectively suppressing pressure fluctuations. Actual measurements show that after using this method, when the valve opening is adjusted significantly (for example, the opening changes by more than 30%), the system pressure fluctuation amplitude is controlled within 2%, which is significantly lower than the 5% fluctuation amplitude of the traditional PID control method, achieving smooth transition control.

[0070] In an alternative embodiment, a dynamic opening adjustment compensation signal is sent to the first bypass valve and the second bypass valve, so that the opening of the first bypass valve and the second bypass valve is gradually adjusted to the target opening combination, comprising: According to the proportional gain coefficient, the differential gain coefficient and the fluctuation suppression coefficient, a first compensation signal and a second compensation signal are generated; the state vector and the first compensation signal are input into an evaluation network to obtain a first state value, and the state vector and the second compensation signal are input into the evaluation network to obtain a second state value; According to the first state value and the second state value, a reward value is calculated, and the state vector, the first compensation signal, the second compensation signal, the reward value and the next time state vector are stored in an experience pool; The execution network and the evaluation network are optimized and trained by obtaining samples from the experience pool; the first compensation signal and the second compensation signal are smoothed to obtain a smoothed compensation signal; and the opening of the first bypass valve and the opening of the second bypass valve are adjusted according to the smoothed compensation signal.

[0071] In this embodiment, the process of sending a dynamic opening adjustment compensation signal to the first bypass valve and the second bypass valve is as follows. The system first generates a first compensation signal and a second compensation signal according to the proportional gain coefficient, the differential gain coefficient and the fluctuation suppression coefficient. In actual application, the proportional gain coefficient can be set to 0.8, the differential gain coefficient to 0.5, and the fluctuation suppression coefficient to 0.3. The system uses a deep reinforcement learning method to generate the compensation signal, wherein the first compensation signal is for the first bypass valve and the second compensation signal is for the second bypass valve. When the system detects that the current opening of the first bypass valve is 40% and the target opening is 60%, the first compensation signal generated based on the above parameters may be +5%, indicating that the opening needs to be increased; when the current opening of the second bypass valve is 70% and the target opening is 50%, the generated second compensation signal may be -4%, indicating that the opening needs to be reduced.

[0072] After the compensation signal is generated, the system inputs the state vector and the first compensation signal into the evaluation network to obtain a first state value, and inputs the state vector and the second compensation signal into the evaluation network to obtain a second state value. The state vector contains key parameters of the current system operation, such as the opening values of the current two bypass valves, fluid pressure values, temperature values, etc. Assuming that the current state vector is [40%, 70%, 0.8 MPa, 85°C], indicating that the first bypass valve opening is 40%, the second bypass valve opening is 70%, the system pressure is 0.8 MPa, and the temperature is 85°C. The evaluation network is a neural network containing three hidden layers, with node numbers of 64, 32, and 16 respectively. After inputting the state vector and the first compensation signal [40%, 70%, 0.8 MPa, 85°C, +5%] into the evaluation network, the first state value 0.82 is calculated; after inputting the state vector and the second compensation signal [40%, 70%, 0.8 MPa, 85°C, -4%] into the evaluation network, the second state value 0.76 is obtained.

[0073] The system calculates the reward value according to the first state value and the second state value. The calculation of the reward value considers two aspects: one is the closeness of the current opening to the target opening, and the other is the stability of the system. In this example, the reward value can be represented as the weighted sum of the first state value and the second state value minus the fluctuation penalty term. Assuming that the first state value weight is 0.6, the second state value weight is 0.4, and the fluctuation penalty term is 0.1, the reward value is calculated as 0.6x0.82+0.4x0.76-0.1=0.792. Subsequently, the system stores the state vector [40%, 70%, 0.8 MPa, 85°C], the first compensation signal +5%, the second compensation signal -4%, the reward value 0.792, and the next time state vector [45%, 66%, 0.82 MPa, 84°C] as a group of experience data into the experience pool.

[0074] The experience pool uses a first-in-first-out data structure, and the capacity is set to 10,000 groups of experience data. When new data is added, if the experience pool is full, the earliest stored data will be removed. The system randomly extracts a batch of samples (such as 256 groups) from the experience pool to optimize and train the execution network and the evaluation network. The training process uses the backpropagation algorithm to update the network weights, with a learning rate of 0.001 and a training iteration number of 100. Through continuous training, the execution network and the evaluation network can gradually improve the ability to generate appropriate compensation signals.

[0075] In order to avoid the impact of the mutation of the compensation signal on the system, the system smoothes the first compensation signal and the second compensation signal to obtain a smoothed compensation signal. The smoothing adopts an exponential weighted moving average method, and the smoothing factor is set to 0.7. For example, if the original first compensation signal at the current time is +5%, and the smoothed compensation signal at the last time is +3%, then the smoothed first compensation signal this time is 0.7*5%+0.3*3%=4.4%. Similarly, if the original second compensation signal at the current time is -4%, and the smoothed compensation signal at the last time is -2%, then the smoothed second compensation signal this time is 0.7*(-4%)+0.3*(-2%)=-3.4%.

[0076] The system adjusts the opening degree of the first bypass valve and the second bypass valve according to the smoothed compensation signal. The adjustment process adopts a step-by-step adjustment strategy, and the single adjustment amplitude does not exceed 5% to ensure smooth transition of the system. The system performs adjustment every 100 milliseconds until the opening degrees of the two bypass valves reach the target opening degree combination. In actual application, if the initial opening degree of the first bypass valve is 40%, the target opening degree is 60%, the opening degree of the second bypass valve is 70%, and the target opening degree is 50%, then about 4-5 seconds of adjustment process may be required, and the opening degree of the first bypass valve gradually increases (40%→44.4%→48.8%→53.2%→57.6%→60%), and the opening degree of the second bypass valve gradually decreases (70%→66.6%→63.2%→59.8%→56.4%→53.0%→50%).

[0077] Through the above method, the system can realize intelligent dynamic adjustment of the opening degree of the bypass valve based on deep reinforcement learning technology, which not only ensures the accuracy of the adjustment, but also ensures the stability of the adjustment process, effectively avoids the oscillation and over-adjustment phenomenon that may occur in traditional PID control, and improves the stability and response speed of the fluid system. Experiments show that, compared with the traditional control method, this method can reduce the steady-state error from ±3% to ±0.8%, and shorten the adjustment time by 20%.

[0078] The hydraulic cooler double-bypass valve pressure self-adaptive adjustment system of the embodiment of the application comprises: The first unit is configured to read pressure data from a data buffer of the hydraulic cooler, calculate a current pressure fluctuation trend characteristic value, perform characteristic matching between the current pressure fluctuation trend characteristic value and pressure fluctuation characteristic values in historical data, and generate a pressure adjustment initial instruction; and send opening degree adjustment signals to the first bypass valve and the second bypass valve according to the pressure adjustment initial instruction, and start a pressure response monitoring timer. The second unit is configured to collect temperature sensor data and flow sensor data in the hydraulic cooler in real time during a monitoring period of the pressure response monitoring timer, calculate a current hydraulic oil viscosity value according to the temperature sensor data, and calculate a Reynolds number of each pipeline section in combination with the flow sensor data; when the Reynolds number of any pipeline section exceeds a flow state transition threshold value, enter a flow state transition control mode, calculate an optimal flow rate based on a pressure gradient of an upstream and downstream of the pipeline section, and convert the optimal flow rate into a target opening degree combination of the first bypass valve and the second bypass valve; The third unit is configured to send an opening degree dynamic adjustment compensation signal to the first bypass valve and the second bypass valve, so that the opening degrees of the first bypass valve and the second bypass valve are gradually adjusted to the target opening degree combination; during the adjustment process, a pressure fluctuation rate is calculated in real time according to an opening degree error change rate, and a network output gain coefficient is executed; when the pressure fluctuation rate is lower than a steady-state fluctuation threshold value for a continuous period, a current first bypass valve opening degree value and a current second bypass valve opening degree value are stored in a steady-state working condition database; and a mapping relationship between the opening degree values and the pressure response is established based on the opening degree values in the steady-state working condition database, for opening degree prediction in the next round of pressure adjustment.

[0079] In a third aspect, an electronic device is provided, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0080] In a fourth aspect, a computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0081] The present application can be a method, device, system or computer program product. The computer program product can include a computer readable storage medium having computer readable program instructions stored therein, which are used to perform various aspects of the present application.

[0082] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for adaptively adjusting the pressure of a double bypass valve of a hydraulic cooler, characterized in that: include: Reading pressure data from a data buffer of the hydraulic cooler, calculating a current pressure fluctuation trend characteristic value, and performing feature matching between the current pressure fluctuation trend characteristic value and pressure fluctuation characteristic values ​​in historical data to generate an initial pressure adjustment instruction; sending opening adjustment signals to the first bypass valve and the second bypass valve respectively based on the initial pressure adjustment instruction, and simultaneously starting a pressure response monitoring timer; During a monitoring period of the pressure response monitoring timer, temperature sensor data and flow sensor data in the hydraulic cooler are collected in real time, a current hydraulic oil viscosity value is calculated based on the temperature sensor data, and the Reynolds number of each pipeline section is calculated in combination with the flow sensor data; when the Reynolds number of any pipeline section exceeds a flow regime transition threshold, a flow regime transition control mode is entered, an optimal flow rate is calculated based on the pressure gradient upstream and downstream of the pipeline section, and the optimal flow rate is converted into a target opening combination for the first bypass valve and the second bypass valve; Sending a dynamic opening adjustment compensation signal to the first bypass valve and the second bypass valve to gradually adjust their openings to the target opening combination; During the adjustment process, the pressure fluctuation rate is calculated in real time based on the opening error change rate, and a gain coefficient is output through the execution network; when the pressure fluctuation rate is lower than the steady-state fluctuation threshold for one consecutive cycle, the current first bypass valve opening value and the current second bypass valve opening value are stored in the steady-state operating condition database; Based on the opening values ​​in the steady-state operating condition database, a mapping relationship between the opening value and the pressure response is established for the opening prediction in the next round of pressure regulation.

2. The method according to claim 1, characterized in that Read the pressure data from the data buffer of the hydraulic cooler, calculate the current pressure fluctuation trend characteristic value, and perform feature matching between the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic value in the historical data to generate the initial pressure regulation instruction, including: Obtaining pressure data of the hydraulic cooler, and performing weighted summation of the pressure standard deviation, pressure skewness coefficient, and pressure kurtosis coefficient of the pressure data as a current pressure fluctuation trend characteristic value; Calculating the Mahalanobis distance between the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic value in the historical data, multiplying the difference vector between the current pressure fluctuation trend characteristic value and the pressure fluctuation characteristic value in the historical data by the inverse matrix of the characteristic covariance matrix, the Mahalanobis distance being the square root of the product; selecting a historical adjustment instruction corresponding to a pressure fluctuation characteristic value in the historical data with the smallest Mahalanobis distance, and multiplying the historical adjustment instruction by a correction coefficient to obtain a current adjustment instruction, wherein the correction coefficient is formed by the sum of the product of the Mahalanobis distance divided by the distance reference value and the compensation factor, and 1; The current adjustment instruction is executed to adjust the pressure of the hydraulic cooler.

3. The method according to claim 1, characterized in that When the Reynolds number of any pipeline section exceeds the flow transition threshold, the flow transition control mode is entered, and the optimal flow rate is calculated based on the pressure gradient upstream and downstream of the pipeline section, and the optimal flow rate is converted into a target opening combination of the first bypass valve and the second bypass valve, including: Based on the flow rate, pipe diameter, hydraulic oil density, and dynamic viscosity at a reference temperature of the hydraulic cooler pipe, a temperature correction coefficient is calculated according to the difference between the current temperature and the reference temperature, the dynamic viscosity at the current temperature is obtained using the temperature correction coefficient, and the Reynolds number of the pipe is obtained by dividing the product of the hydraulic oil density, flow rate, and pipe diameter by the dynamic viscosity at the current temperature; Obtaining a local pressure gradient of the pipeline, dividing the Reynolds number by a critical Reynolds number to obtain a first ratio, dividing the absolute value of the local pressure gradient by the critical pressure gradient to obtain a second ratio, and using a weighted sum of the first ratio and the second ratio as a flow regime transition criterion; When the flow transition criterion is greater than 1, the target flow rate at which energy loss is minimized is calculated based on the along-the-line loss coefficient, the local loss coefficient, and the length of the pipe section of the pipeline; constructing an optimization objective function for the opening of the first bypass valve and the opening of the second bypass valve based on the target flow rate, flow coefficient, hydraulic oil density, and flow area, wherein the optimization objective function includes a flow deviation term, an opening difference term, and an opening change rate term; and solving the optimization objective function to obtain the target opening of the first bypass valve and the target opening of the second bypass valve; A time constant is dynamically calculated based on a deviation between the Reynolds number and the critical Reynolds number, and the first bypass valve target opening and the second bypass valve target opening are progressively adjusted according to the time constant to achieve a smooth transition of the flow state.

4. The method according to claim 1, wherein Storing the current first bypass valve opening value and the current second bypass valve opening value in a steady-state operating condition database; Based on the opening value in the steady-state operating condition database, a mapping relationship between the opening value and the pressure response is established for the opening prediction in the next round of pressure regulation, including: Acquire multiple sets of training samples of the first bypass valve opening, the second bypass valve opening, and the system pressure from a steady-state operating condition database; Based on the training samples, a nonlinear mapping model of opening and pressure is established; a basis function center and an expansion parameter are determined, and a radial basis function value of each group of training samples is calculated based on the first bypass valve opening and the second bypass valve opening; wherein the radial basis function value is obtained by dividing the sum of the squares of the differences between the first bypass valve opening and the second bypass valve opening and the basis function center by the square of the expansion parameter, and taking the negative exponent of the quotient; Accumulating the product of the radial basis function value and the weight coefficient to obtain a nonlinear mapping function; calculating the sum of the squares of the differences between the output value of the nonlinear mapping function and the system pressure, calculating the sum of the squares of the weight coefficients, and taking the weighted sum of the sum of the squares of the differences and the sum of the squares of the weight coefficients as the loss function; Obtaining a target pressure value, and using a difference between the target pressure value and an output value of the nonlinear mapping function as an objective function; calculating a partial derivative of the objective function with respect to the first bypass valve opening and a partial derivative of the objective function with respect to the second bypass valve opening; The first bypass valve opening and the second bypass valve opening are iteratively updated according to the partial derivative of the first bypass valve opening and the partial derivative of the second bypass valve opening to obtain a first bypass valve target opening and a second bypass valve target opening.

5. The method according to claim 1, wherein Sending a dynamic opening adjustment compensation signal to the first bypass valve and the second bypass valve to gradually adjust their openings to the target opening combination; During the adjustment process, the pressure fluctuation rate is calculated in real time based on the opening error change rate, and the gain coefficient is output through the execution network, including: Acquire a system pressure signal sequence, a first bypass valve opening sequence, and a second bypass valve opening sequence, and construct the system pressure signal sequence, the first bypass valve opening sequence, and the second bypass valve opening sequence into a historical data sequence; Calculating a system pressure fluctuation rate, obtaining a difference between a target opening of the first bypass valve and a current opening of the first bypass valve to obtain a first opening error, obtaining a difference between a target opening of the second bypass valve and a current opening of the second bypass valve to obtain a second opening error, and calculating a change rate of the first opening error and a change rate of the second opening error; constructing the system pressure fluctuation rate, the first opening error, the second opening error, the change rate of the first opening error, the change rate of the second opening error, and the historical data sequence into a state vector; The state vector is input into the execution network, and the execution network outputs a proportional gain coefficient, a differential gain coefficient and a fluctuation suppression coefficient; wherein the proportional gain coefficient has a negative exponential relationship with the system pressure fluctuation rate, the differential gain coefficient has a negative exponential relationship with the system pressure fluctuation rate, and the fluctuation suppression coefficient has a positive correlation with the system pressure fluctuation rate.

6. The method according to claim 5, characterized in that Sending a dynamic opening adjustment compensation signal to the first bypass valve and the second bypass valve to gradually adjust their openings to the target opening combination includes: generating a first compensation signal and a second compensation signal according to the proportional gain coefficient, the differential gain coefficient, and the fluctuation suppression coefficient; inputting the state vector and the first compensation signal into an evaluation network to obtain a first state value, and inputting the state vector and the second compensation signal into the evaluation network to obtain a second state value; Calculating a reward value according to the first state value and the second state value, and storing the state vector, the first compensation signal, the second compensation signal, the reward value, and the next-moment state vector into an experience pool; Obtain samples from the experience pool to optimize and train the execution network and the evaluation network; smooth the first compensation signal and the second compensation signal to obtain a smoothed compensation signal; and adjust the opening of the first bypass valve and the opening of the second bypass valve according to the smoothed compensation signal.

7. A hydraulic cooler dual bypass valve pressure adaptive regulation system, used to implement the method according to any one of claims 1 to 6, characterized in that: include: The first unit is configured to read pressure data from a data buffer of the hydraulic cooler, calculate a current pressure fluctuation trend characteristic value, perform feature matching on the current pressure fluctuation trend characteristic value with pressure fluctuation characteristic values ​​in historical data, and generate an initial pressure adjustment instruction; send an opening adjustment signal to the first bypass valve and the second bypass valve respectively based on the initial pressure adjustment instruction, and simultaneously start a pressure response monitoring timer; a second unit configured to collect temperature sensor data and flow sensor data in the hydraulic cooler in real time during a monitoring period of the pressure response monitoring timer, calculate a current hydraulic oil viscosity value based on the temperature sensor data, and calculate a Reynolds number for each pipeline section in combination with the flow sensor data; enter a flow transition control mode when the Reynolds number for any pipeline section exceeds a flow transition threshold, calculate an optimal flow rate based on a pressure gradient upstream and downstream of the pipeline section, and convert the optimal flow rate into a target opening combination for the first bypass valve and the second bypass valve; a third unit, configured to send an opening dynamic adjustment compensation signal to the first bypass valve and the second bypass valve, so as to gradually adjust their openings to the target opening combination; During the adjustment process, the pressure fluctuation rate is calculated in real time based on the opening error change rate, and a gain coefficient is output through the execution network; when the pressure fluctuation rate is lower than the steady-state fluctuation threshold for one consecutive cycle, the current first bypass valve opening value and the current second bypass valve opening value are stored in the steady-state operating condition database; Based on the opening values ​​in the steady-state operating condition database, a mapping relationship between the opening value and the pressure response is established for the opening prediction in the next round of pressure regulation.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images