A pressure control method, system, and storage medium

Abstract

This invention provides a pressure control method, system, and storage medium. The method includes: acquiring the previous round's actual pressure and the corresponding previous round's preset pressure; determining the previous round's actual pressure error and the corresponding rate of change of the previous round's actual error between the previous round's actual pressure and the preset pressure; performing fuzzification processing on the previous round's actual pressure error and the rate of change of the previous round's actual error to obtain fuzzy previous round pressure information; acquiring a fuzzy control rule base; substituting the fuzzy previous round pressure information into the fuzzy control rule base to obtain initial fuzzy control information; performing defuzzification processing on the initial fuzzy control information to obtain initial defuzzified control information; acquiring the previous round's external information corresponding to the output previous round's actual pressure; predicting the current round's external information for control based on the previous round's external information; adjusting the initial defuzzified control information based on the current round's external information to obtain target control information; and using the target control information for pressure control. This application provides precise and stable pressure control.

Description

Technical Field

[0001] This application relates to the field of pressure technology, and in particular to a pressure control method, system and storage medium. Background Technology

[0002] In many fields such as pressure instrument calibration, chemical production, and laboratory testing, there is a need for pressure that can generate high precision and stability. The accuracy and stability of this pressure directly determine the reliability of calibration results and the quality of the production process. However, the pressure generation process is affected by various factors, such as ambient temperature, media characteristics, and equipment aging, which cause the pressure generation system to exhibit nonlinearity and time-varying characteristics, making precise pressure generation difficult.

[0003] Currently, the traditional method for achieving automatic pressure generation mainly relies on the classic PID control algorithm. PID control regulates pressure through three stages: proportional, integral, and derivative, adjusting the control input of the equipment based on the pressure error. However, because PID control typically requires a precise mathematical model, and the nonlinearity and time-varying nature of the pressure generation process make it difficult to establish an accurate model, the parameter tuning of PID control is challenging, thus making it difficult to guarantee accurate and stable pressure control. Summary of the Invention

[0004] To accurately and stably control pressure, embodiments of this application provide a pressure control method, system, and storage medium.

[0005] In a first aspect, this embodiment provides a pressure control method, the method comprising: Obtain the previous round's actual pressure and the corresponding previous round's preset pressure, determine the error between the previous round's actual pressure and the previous round's preset pressure, and the rate of change of the previous round's actual pressure error corresponding to the previous round's actual pressure error; The fuzzy upper wheel pressure information is obtained by fuzzifying the actual pressure error of the upper wheel and the rate of change of the actual error of the upper wheel. Obtain the fuzzy control rule base, substitute the fuzzy upper wheel pressure information into the fuzzy control rule base to obtain the initial fuzzy control information, and perform defuzzification processing on the initial fuzzy control information to obtain the initial defuzzy control information; The system acquires and outputs the external information of the previous wheel corresponding to the actual pressure of the previous wheel, predicts the external information of the current wheel for control based on the external information of the previous wheel, adjusts the initial unfuzzy control information based on the external information of the current wheel to obtain the target control information, and uses the target control information for pressure control.

[0006] In some embodiments, the blurred upper wheel pressure information includes blurred pressure error and blurred error change rate, and the process of blurring the actual upper wheel pressure error and the actual upper wheel error change rate to obtain blurred upper wheel pressure information includes: Obtain the error fuzzy set and the first error membership function corresponding to the pressure error attribute, and the change rate fuzzy set and the first error change rate membership function corresponding to the pressure error change rate attribute; The first error membership function is used to process the actual pressure error of the upper wheel to obtain the first error membership information, wherein the first error membership information includes several first error membership sub-information; Obtain the first reference error membership sub-information representing the high strength of membership in the first error membership information, determine whether there is a strong correlation between the first reference error membership sub-information, if not, substitute the highest strength first reference error membership sub-information in the first error membership information into the error fuzzy set to obtain the fuzzy pressure error; If so, obtain the second error membership function corresponding to the pressure error attribute, use the second error membership function to process the actual pressure error of the previous round to obtain the second error membership information, and determine the fuzzy pressure error based on the first error membership information and the second error membership information; The first error change rate membership function is used to process the actual error change rate of the previous round to obtain the first error change rate membership information, wherein the first error change rate membership information includes several first error change rate membership sub-information. Obtain the first reference error change rate sub-information representing the high intensity of the first error change rate membership information, determine whether there is a strong correlation between the first reference error change rate sub-information, if not, substitute the highest intensity first reference error change rate sub-information in the first error change rate membership information into the change rate fuzzy set to obtain the fuzzy error change rate. If so, obtain the second error change rate membership function corresponding to the pressure error change rate attribute, use the second error change rate membership function to process the previous round's actual error change rate to obtain the second error change rate membership information, and determine the fuzzy error change rate based on the first error change rate membership information and the second error change rate membership information.

[0007] In some embodiments, the second error membership information includes several second error membership sub-information, the error fuzzy set includes several error fuzzy languages, each error membership sub-information corresponds to one error fuzzy language, and determining the fuzzy pressure error based on the first error membership information and the second error membership information includes: Obtain the first weight corresponding to the first error membership function, and multiply the first weight by each first error membership sub-information to obtain the first discrimination value corresponding to each first error membership sub-information; Obtain the second weight corresponding to the second error membership function, and multiply the second weight by each second error membership sub-information to obtain the second discrimination value corresponding to each second error membership sub-information; The first and second distinction values ​​corresponding to the same error fuzzy language are added together to obtain the distinction value of the error fuzzy language, and the error fuzzy language corresponding to the largest distinction value among all distinction values ​​is determined as the fuzzy pressure error.

[0008] In some embodiments, obtaining the fuzzy control rule base includes: Acquire historical fuzzy pressure information and corresponding historical fuzzy control information that have met the standards in the completed pressure control work; The corresponding historical fuzzy pressure information and historical fuzzy control information are divided into training information group, verification information group and test information group; The initial neural network is trained using the training information set to obtain the model to be validated. The verification information group is used to verify the model to be verified, and the model to be verified is adjusted to obtain the model to be used. The test information group is used to test the model to be used to obtain test results. The test results are then used to determine whether the test results are qualified. If they are qualified, the model to be used is determined as a fuzzy control rule library. If it fails, the corresponding historical fuzzy pressure information and historical fuzzy control information are reclassified into training information group, verification information group and test information group.

[0009] In some embodiments, the process of defuzzifying the initial fuzzy control information to obtain the initial defuzzified control information includes: Obtain the set of fuzzy control quantities and the membership function of the control quantities corresponding to the initial fuzzy control information, wherein the fuzzy set of control quantities includes several fuzzy linguistic terms that characterize different control intensities, and each fuzzy linguistic term corresponds to a membership distribution interval in the membership function of the control quantities; Obtain the membership degree value corresponding to each fuzzy linguistic item in the initial fuzzy control information, and establish the relationship between the fuzzy linguistic item and the membership degree value based on the membership degree value; Based on the membership function of the control quantity, the universe center value corresponding to each fuzzy linguistic item is determined, and the membership value and the universe center value are processed by the weighted average method to obtain the initial solution fuzzy control information.

[0010] In some embodiments, predicting the current-round external information for control based on the previous-round external information includes: using a BP neural network to process the previous-round external information to predict the current-round external information for control.

[0011] In some embodiments, adjusting the initial unfuzzy control information based on the current external information to obtain the target control information includes: Substitute the current external information into a preset model corresponding to external information and pressure effects to obtain pressure effect information; The pressure influence information is substituted into a preset pressure-valve control correspondence to obtain control adjustment information; The target control information is obtained by superimposing the control adjustment information onto the initial unfuzzy control information.

[0012] In some embodiments, obtaining the target control information further includes: Obtain the actual pressure of the current round output corresponding to the target control information and the preset pressure of the current round corresponding to the actual pressure of the current round, and determine the actual pressure error of the current round between the actual pressure of the current round and the preset pressure of the current round; Determine whether the actual pressure error of this round meets the standard. If it does, proceed to the next round of pressure control. If the target is not met, obtain the external information corresponding to the actual pressure of this round, and adjust the output layer weights of the BP neural network based on the external information of this round and the real-time external information to obtain a new deep learning model.

[0013] Secondly, this embodiment provides a pressure control system, the system comprising: a pressure detection module, a pressure processing module, and a pressure control module; wherein, The pressure detection module is used to obtain the previous wheel actual pressure output and the previous wheel preset pressure corresponding to the previous wheel actual pressure, and to determine the previous wheel actual pressure error between the previous wheel actual pressure and the previous wheel preset pressure and the previous wheel actual error change rate corresponding to the previous wheel actual pressure error. The pressure processing module is used to perform fuzzification processing on the actual pressure error of the upper wheel and the change rate of the actual error of the upper wheel to obtain fuzzy upper wheel pressure information; obtain a fuzzy control rule base, substitute the fuzzy upper wheel pressure information into the fuzzy control rule base to obtain initial fuzzy control information, and perform defuzzification processing on the initial fuzzy control information to obtain initial defuzzy control information. The pressure control module is used to acquire the external information of the previous wheel corresponding to the actual pressure of the previous wheel, predict the external information of the current wheel based on the external information of the previous wheel, adjust the initial defuzzification control information based on the external information of the current wheel to obtain target control information, and use the target control information to perform pressure control.

[0014] Thirdly, this embodiment provides a computer-readable storage medium having a computer program stored thereon that can run on a processor, wherein when the computer program is executed by the processor, it implements a pressure control method as described in the first aspect.

[0015] By employing the above method, this application first obtains the previous round's actual pressure and the corresponding previous round's preset pressure, and determines the previous round's actual pressure error between the previous round's actual pressure and the previous round's preset pressure, as well as the rate of change of the previous round's actual error corresponding to the previous round's actual pressure error. By selecting pressure error and error change rate as input variables for the pressure control method, the current state and changing trend of the pressure generation system can be more comprehensively reflected. Then, the previous round's actual pressure error and the previous round's actual error change rate are fuzzified to obtain fuzzy previous round pressure information. Next, a fuzzy control rule base is obtained, and the fuzzy previous round pressure information is substituted into the fuzzy control rule base to obtain initial fuzzy control information. The initial fuzzy control information is then defuzzified to obtain initial defuzzified control information. Finally, the previous round's external information corresponding to the output of the previous round's actual pressure is obtained. Based on the previous round's external information, the current round's external information for control is predicted. Based on the current round's external information, the initial defuzzified control information is adjusted to obtain target control information, and the target control information is used for pressure control. By using fuzzy control, without the need for a precise mathematical model, it can adapt to the nonlinearity and time-varying nature of pressure-generating systems. At the same time, it also takes into account external information to further adjust the fuzzy control, thereby achieving precise and stable control of pressure and improving the robustness and adaptability of the system. Attached Figure Description

[0016] Figure 1 This is a block diagram of a pressure control method provided in this application.

[0017] Figure 2 This is a flowchart of a method provided in this application for fuzzing the actual pressure error and the rate of change of the actual pressure error of the previous wheel to obtain fuzzed pressure information of the previous wheel.

[0018] Figure 3 This is a block diagram of a method provided in this application for obtaining initial defuzzified control information by defuzzifying initial fuzzy control information.

[0019] Figure 4 This is a connection diagram of a pressure control system provided in this application. Detailed Implementation

[0020] To better understand the purpose, technical solutions, and advantages of this application, it has been described and illustrated below with reference to the accompanying drawings and embodiments. However, those skilled in the art should understand that this application can be implemented without these details. It will be apparent to those skilled in the art that various modifications can be made to the embodiments disclosed in this application, and the general principles defined in this application can be applied to other embodiments and application scenarios without departing from the principles and scope of this application. Therefore, this application is not limited to the illustrated embodiments, but is consistent with the broadest scope claimed in this application.

[0021] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0022] This application relates to pressure control in a pressure generation system. The pressure generation system includes a pressure sensor, a controller, and an actuator. The sensor is connected to the controller via a signal line, and the controller is connected to the actuator via a control line. The entire system has a compact layout, reliable connections between components, and is easy to install and maintain. The pressure control method provided in this application is described from the controller's perspective. Figure 1 This is a block diagram of a pressure control method provided in this application. Figure 1 As shown, a pressure control method includes the following steps: Step S100: Obtain the previous round's actual pressure and the corresponding previous round's preset pressure, and determine the previous round's actual pressure error between the previous round's actual pressure and the previous round's preset pressure, as well as the previous round's actual error change rate corresponding to the previous round's actual pressure error.

[0023] The "previous round actual pressure" mentioned above refers to the most recently detected actual pressure value by the pressure sensor, while the "previous round preset pressure" refers to the pressure value expected to be generated by the pressure generation system. The previous round actual pressure can be obtained by reading the pressure sensor data, and the previous round preset pressure can be obtained by reading information stored in the control terminal. The control terminal stores the pressure values ​​generated by the expected pressure generation system each time.

[0024] Next, subtracting the preset pressure from the previous pressure yields the error in the previous pressure, representing the error in the most recent pressure generation by the pressure generation system. Then, subtracting the actual pressure of the previous pressure (the pressure of the pressure two years prior) from the previous actual pressure, and dividing by the time elapsed between obtaining the previous pressure error and the previous pressure error, yields the rate of change of the previous pressure error. By selecting pressure error and its rate of change as input variables for the pressure control method, a more comprehensive reflection of the current state and trends of the pressure generation system can be achieved.

[0025] Step S200: The actual pressure error and the rate of change of the actual pressure error of the previous wheel are fuzzed to obtain the fuzzed pressure information of the previous wheel.

[0026] Once the input variables for the pressure control method are determined, corresponding processing is performed on them. Among these, the fuzzified pressure information includes the fuzzified pressure error and the fuzzified error change rate. The fuzzified pressure error corresponds to the information obtained by fuzzifying the actual pressure error of the previous wheel, and the fuzzified error change rate corresponds to the information obtained by fuzzifying the aforementioned actual error change rate. Figure 2 This is a flowchart illustrating the method provided in this application for fuzzifying the actual pressure error and the rate of change of the actual pressure error of the previous wheel to obtain fuzzified pressure information of the previous wheel. (For example...) Figure 2 As shown, the process of fuzzifying the actual pressure error and the rate of change of the actual pressure error of the previous wheel to obtain the fuzzified pressure information of the previous wheel includes the following steps: Step S201: Obtain the error fuzzy set and the first error membership function corresponding to the pressure error attribute, and the change rate fuzzy set and the first error change rate membership function corresponding to the pressure error change rate attribute.

[0027] Step S202: The first error membership function is used to process the actual pressure error of the previous round to obtain the first error membership information, wherein the first error membership information includes several first error membership sub-information.

[0028] Step S203: Obtain the first reference error sub-information representing the high intensity of membership in the first error membership information, determine whether there is a strong correlation between the first reference error sub-information, if not, substitute the highest intensity first reference error sub-information in the first error membership information into the error fuzzy set to obtain the fuzzy pressure error.

[0029] Step S204: If yes, obtain the second error membership function corresponding to the pressure error attribute, use the second error membership function to process the actual pressure error of the previous round to obtain the second error membership information, and determine the fuzzy pressure error based on the first error membership information and the second error membership information.

[0030] Step S205: Process the actual error change rate of the previous round using the first error change rate membership function to obtain the first error change rate membership information, wherein the first error change rate membership information includes several first error change rate membership sub-information.

[0031] Step S206: Obtain the first reference error change rate sub-information representing the high intensity of the first error change rate membership information, determine whether there is a strong correlation between the first reference error change rate sub-information, if not, substitute the highest intensity first reference error change rate sub-information in the first error change rate membership information into the change rate fuzzy set to obtain the fuzzy error change rate.

[0032] Step S207: If yes, obtain the second error change rate membership function corresponding to the pressure error change rate attribute, use the second error change rate membership function to process the previous round's actual error change rate to obtain the second error change rate membership information, and determine the fuzzy error change rate based on the first error change rate membership information and the second error change rate membership information.

[0033] The aforementioned fuzzy error set is determined by the staff based on the actual application scenario. Different application scenarios have different acceptable levels of error granularity, but the fuzzy error set must include at least negative, zero, and positive values. This application preferably defines the fuzzy error set as {negative large, negative small, zero, positive small, positive large}. The aforementioned first error membership function is a membership function selected by the staff based on actual needs. Considering the relatively simple nature of the error, this application selects a triangular membership function as the first error membership function. Both the fuzzy error set and the first error membership function are pre-stored in the controller. The fuzzy error set and the first error membership function corresponding to the pressure error attribute can be obtained by viewing the stored information.

[0034] Similarly, the aforementioned fuzzy set of rate of change is determined by the staff with reference to the aforementioned fuzzy set of error; that is, this application preferably defines the fuzzy set of rate of change as {negative large, negative small, zero, positive small, positive large}. The aforementioned membership function of the first error rate of change is also determined by the staff with reference to the aforementioned first error membership function; that is, this application selects the triangular membership function as the membership function of the first error rate of change. The fuzzy set of rate of change and the membership function of the first error rate of change are also pre-stored in the controller, and the fuzzy set of rate of change and the membership function of the first error rate of change corresponding to the pressure error rate of change attribute can be obtained by viewing the stored information.

[0035] The error fuzzy set includes several error fuzzy languages. The fuzzification process for the previous round's actual pressure error yields the fuzzy pressure error. First, the previous round's actual pressure error is processed using the first error membership function, resulting in a probability value for each error fuzzy language in the error fuzzy set. Each error fuzzy language corresponds to an error membership sub-information. Integrating the corresponding error fuzzy language and probability value gives the error membership sub-information for that fuzzy language. For example, if the probability value for the error fuzzy language "negative small" is 0.4, then the error membership sub-information for "negative small" is that the pressure error belongs to the negative small category with a probability of 0.4. The error membership sub-information for each error fuzzy language constitutes the first error membership information.

[0036] The aforementioned high membership strength refers to the probability that is greater than a preset probability among all first error membership sub-information. The preset probability can be determined based on actual conditions, ensuring that at least two first error membership sub-information have a probability value greater than this preset probability. Here, the magnitude of the preset probability is not further limited. By comparing the probability corresponding to each reference error membership sub-information included in the first error membership information with the preset probability, the first error membership sub-information corresponding to probabilities greater than the preset probability is determined as the first reference error membership sub-information.

[0037] The aforementioned strong correlation refers to a situation where the difference between the probabilities corresponding to the first reference error's subordinate information is not less than a preset probability difference. This preset probability difference is determined based on actual circumstances, and no further limitations are imposed here. By comparing the pairwise subtraction of all probabilities corresponding to the first reference error's subordinate information with the preset probability difference, if the pairwise subtraction is not less than the preset probability difference, then the first reference error's subordinate information is not strongly correlated. If the pairwise subtraction of probabilities is less than the preset probability difference, then the first reference error's subordinate information is strongly correlated.

[0038] The fact that there is no strong correlation between the membership information of the first reference error indicates that the information gap after fuzzification is large. The membership information of the first reference error with the highest probability in the membership information of the first error is directly substituted into the error fuzzy set to obtain the corresponding error fuzzy language, and thus this error fuzzy language is determined as the pressure error of the model.

[0039] The strong correlation between the first reference error membership sub-information indicates a small information gap after fuzzification. Considering that the error fuzzy language corresponding to the first reference error membership sub-information with smaller gaps but not the highest probability might be more suitable as the current fuzzified pressure error, the second error membership function corresponding to the pressure error attribute is then obtained from the stored information. This second error membership function differs from the first error membership function; it can be a trapezoidal or Gaussian membership function. The second error membership function is then used to reprocess the previous round's actual pressure error, obtaining the probability value corresponding to each error fuzzy language in the error fuzzy set. Each error fuzzy language corresponds to one error membership sub-information. Integrating the corresponding error fuzzy language and probability value yields the error membership sub-information corresponding to that error fuzzy language. The error membership sub-information corresponding to each error fuzzy language constitutes the second error membership information. Finally, the first and second error membership information are combined to obtain the fuzzified pressure error of this round of fuzzification. This approach prioritizes using a single error membership function to fuzzify and obtain the fuzzy pressure error. If the fuzzy pressure error cannot be clearly obtained using only a single error membership function, a new error membership function is used to perform fuzzification again. By combining these two fuzzification processes, the final fuzzy pressure error can be determined. This approach can simultaneously balance the efficiency and accuracy of fuzzification, resulting in a more efficient and accurate fuzzy pressure error, which in turn facilitates faster and more precise pressure control in the future.

[0040] The second error membership information includes several second error membership sub-information, each corresponding to an error fuzzy language. Determining the fuzzy pressure error based on the first and second error membership information includes the following steps: Step S204-1: Obtain the first weight corresponding to the first error membership function, and multiply the first weight by each first error membership sub-information to obtain the first discrimination value corresponding to each first error membership sub-information.

[0041] Step S204-2: Obtain the second weight corresponding to the second error membership function, and multiply the second weight by each second error membership sub-information to obtain the second discrimination value corresponding to each second error membership sub-information.

[0042] Step S204-3: Add the first and second distinction values ​​corresponding to the same error fuzzy language to obtain the distinction value of the error fuzzy language, and determine the error fuzzy language corresponding to the largest distinction value among all distinction values ​​as the fuzzy pressure error.

[0043] The first weight mentioned above refers to the ratio of the probability corresponding to each error fuzzy language obtained from the first error membership function, and the second weight refers to the ratio of the probability corresponding to each error fuzzy language obtained from the second error membership function. The values ​​of the first and second weights can be determined according to the actual situation. By multiplying the first weight by the weight in each first error membership sub-information, the first discrimination value corresponding to this first error membership sub-information can be obtained. Similarly, by multiplying the second weight by the weight in each second error membership sub-information, the second discrimination value corresponding to this second error membership sub-information can be obtained.

[0044] Then, the first and second discriminant values ​​corresponding to the same error fuzzy language are added together to obtain the discriminant value of this error fuzzy language. The error fuzzy language corresponding to the largest discriminant value among all obtained discriminant values ​​is then determined as the fuzzy pressure error. If the number of largest discriminant values ​​among all obtained discriminant values ​​is greater than one, then the error fuzzy language corresponding to the one with the highest probability obtained using the first error membership function is determined as the fuzzy pressure error. In this way, the weighted calculation using the dual error membership function achieves accurate correction of the fuzzy pressure error obtained from the fuzzification process, effectively reducing the discrimination of single-function processing.

[0045] Similarly, the fuzzification process for the previous round's actual error change rate yields the fuzzified error change rate. The specific implementation method is the same as the method described above for fuzzifying the previous round's actual pressure error, and will not be repeated here. Again, a single error change rate membership function is used first for fuzzification to obtain the fuzzified error change rate. When a single error change rate membership function alone is insufficient to clearly obtain the fuzzified error change rate, a new error change rate membership function is used for further fuzzification. Combining these two fuzzification processes determines the final fuzzified error change rate, balancing efficiency and accuracy. This results in a more efficient and accurate fuzzified error change rate, facilitating faster and more precise pressure control in the future. Furthermore, weighted calculation using dual error membership functions achieves precise correction of the fuzzified error change rate, effectively reducing the bias inherent in single-function processing.

[0046] Step S300: Obtain the fuzzy control rule base, substitute the fuzzy previous pressure information into the fuzzy control rule base to obtain the initial fuzzy control information, and perform defuzzification processing on the initial fuzzy control information to obtain the initial defuzzy control information.

[0047] The aforementioned fuzzy control rule base refers to the control rules referenced by the initial fuzzy control information obtained based on the fuzzified previous pressure information. Obtaining the fuzzy control rule base includes the following steps: Step S301: Obtain historical fuzzy pressure information and corresponding historical fuzzy control information that have met the standards in the completed pressure control work.

[0048] Step S302: Divide the corresponding historical fuzzy pressure information and historical fuzzy control information into training information group, verification information group and test information group.

[0049] Step S303: Use the training information group to train the initial neural network to obtain the model to be verified.

[0050] Step S304: Use the verification information group to verify the model to be verified in order to adjust the model to be verified and obtain the model to be used.

[0051] Step S305: Use the test information group to test the model to be used and obtain the test results. Determine whether the test results are qualified. If qualified, determine the model to be used as the fuzzy control rule library.

[0052] Step S306: If the result is unqualified, the corresponding historical fuzzy pressure information and historical fuzzy control information are reclassified into training information group, verification information group and test information group.

[0053] The aforementioned model control rule base was established offline and stored in the controller. Specifically, the model control rule base is a deep learning model, which takes fuzzy pressure information as input and outputs fuzzy control information. The deep learning model was trained using a large amount of historical data.

[0054] Specifically, the qualified historical fuzzified pressure information and the corresponding historical fuzzy control information in the completed pressure control work are obtained by reading historical information. Then, the corresponding historical fuzzified pressure information and the historical fuzzy control information are bound together to form input-output information, and a plurality of input-output information pieces can be obtained. Then, the plurality of input-output information pieces are randomly divided into three groups, namely a training information group, a verification information group and a test information group. Wherein, the number of input-output information pieces contained in the training information group accounts for more than half of the total number, and the verification information group and the test information group each account for half of the remaining number of input-output information pieces. Then, the training information group is used to train an untrained initial neural network, and a to-be-verified model is obtained after training is completed. Subsequently, the verification information group is used to verify the to-be-verified model and adjust the to-be-verified model to obtain a to-be-used model. In order to determine whether the to-be-used model meets the use standard, the historical fuzzified pressure information in the test information group is also input into the to-be-used model, a test result is obtained through the output end of the to-be-used model, and a similarity comparison is performed between the test result and the historical fuzzy control information corresponding to the historical fuzzified pressure information. If the similarity is not less than a preset similarity, the test result is qualified; if the similarity is less than the preset similarity, the test result is unqualified. If the test result is determined to be qualified, the obtained to-be-used model is determined as the fuzzy rule control base; if the test result is determined to be unqualified, the corresponding historical fuzzified pressure information and historical fuzzy control information are re-divided into a training information group, a verification information group and a test information group, and the re-determined training information group is used to train the above-obtained to-be-used model, thereby accelerating the progress of determining the fuzzy control rule base. In this way, the fuzzy control rule base is obtained through self-learning based on historical information. Compared with obtaining the rule base by relying on artificial work experience, a more accurate fuzzy control rule base can be obtained, thereby indirectly improving the accuracy of subsequent pressure control.

[0055] After the fuzzified previous-round pressure information and the fuzzy control rule base are obtained, the fuzzified previous-round pressure information is substituted into the fuzzy control rule base to obtain initial fuzzy control information corresponding to the fuzzified previous-round pressure information. In order to enable the actuator to take corresponding measures to realize pressure control, defuzzification processing is also required for the initial fuzzy control information. Figure 3 is a method block diagram of obtaining initial defuzzified control information by performing defuzzification processing on initial fuzzy control information provided by the present application. As Figure 3 shown, obtaining initial defuzzified control information by performing defuzzification processing on initial fuzzy control information includes the following steps: Step S307: Acquire a fuzzy control quantity set and a control quantity membership function corresponding to the initial fuzzy control information, wherein the control quantity fuzzy set includes a plurality of fuzzy language terms characterizing different control intensities, and each fuzzy language term corresponds to one membership distribution interval in the control quantity membership function.

[0056] Step S308: Obtain the membership value corresponding to each fuzzy language item included in the initial fuzzy control information, and establish the relationship between fuzzy language and membership value based on the membership value.

[0057] Step S309: Determine the universe center value corresponding to each fuzzy linguistic term based on the membership function of the control quantity, and use the weighted average method to process the membership value and the universe center value to obtain the initial solution fuzzy control information.

[0058] The aforementioned set of fuzzy control variables was determined by the staff based on actual application scenarios. Different application scenarios have varying degrees of acceptance for the granularity of control division, but the set of fuzzy control variables must at least include increases, constants, and decreases. This application preferably defines the error fuzzy set as {significant decrease, moderate decrease, constant, moderate increase, significant increase}, where each fuzzy linguistic term corresponds to a different control intensity and has a clear control intent and a corresponding universe of discourse. The universe of discourse is represented by quantized values, covering the adjustment range of the execution structure. For example, if the quantization range is set to -50 to 50, the negative sign represents a decrease in adjustment action, which in turn represents an increase in adjustment action. The five fuzzy language terms are: a significant reduction, corresponding to the domain interval [-50, -30], which is intended to significantly reduce the valve opening of the execution structure; a moderate reduction, corresponding to the domain interval [-30, -10], which is intended to moderately reduce the valve opening of the execution structure; a no change, corresponding to the domain interval [-10, 10]; a moderate increase, corresponding to the domain interval [10, 30]; and a significant increase, corresponding to the domain interval [30, 50].

[0059] In selecting the membership function for the control variable, a triangular membership function is adopted, taking into account both computational simplicity and the accuracy of fuzzy partitioning. The membership function for each fuzzy linguistic term is defined by its corresponding feature point. Taking the moderate reduction corresponding to the initial fuzzy control information as an example, its feature points are (-30, 0), (-20, 1), and (-10, 0). This means that when the quantized value of the control variable is -20, the membership degree of the linguistic term reaches its highest value of 1. When the quantized value is in the range of -30 to -20 or -20 to -10, the membership degree increases or decreases linearly with the change in the quantized value. When the quantized value exceeds the range of [-30, -10], the membership degree is zero.

[0060] Due to the fuzzy diffusion effect in the fuzzy inference process, in addition to the initial fuzzy control information being moderately reduced, it is also necessary to extract the membership values ​​of adjacent fuzzy linguistic terms to comprehensively reflect the fuzzy characteristics of the control decision. By tracing the generation logic of fuzzy inference, the membership values ​​corresponding to each fuzzy linguistic term are finally determined. For example, a significant reduction results in 0.1, a moderate reduction in 0.9, no change in 0.2, and a moderate or significant increase in both are 0. Then, based on the determined membership values ​​corresponding to each fuzzy linguistic term, the relationship between the fuzzy linguistic term and the membership value is established, clearly presenting the degree of membership corresponding to different control intensities.

[0061] The center value of the universe of discourse is the typical precise control quantity corresponding to each fuzzy linguistic term in the control quantity universe of discourse, providing the basis for subsequent precise calculations. It is determined by taking the midpoint of the universe of discourse interval for each fuzzy linguistic term. For example, a significantly reduced center value is the center of the interval [-50, -30], a moderately reduced center value is the center of the interval [-30, -10], an unchanged center value is the center of the interval [-10, 10], a moderately increased center value is the center of the interval [10, 30], and a significantly increased center value is the center of the interval [30, 50]. The center value of the universe of discourse corresponding to each fuzzy linguistic term can be obtained through simple calculation.

[0062] Finally, a weighted average method is used for calculation. This involves multiplying the membership values ​​of each fuzzy linguistic term by their corresponding universe center values, summing the results, and then dividing by the sum of the membership values ​​to obtain accurate initial fuzzy control information. The effective range of the control quantity is pre-set based on the physical parameters of the actuator, ensuring that the generated control does not exceed the actuator's operating capacity and reducing the possibility of actuator damage or control failure. This transforms the abstract fuzzy control information obtained from fuzzy inference into precise control parameters that the actuator can directly respond to, building a crucial bridge between fuzzy decision-making and physical execution. The weighted average method integrates information from multiple related fuzzy linguistic terms, reducing information loss during transformation. Combined with the setting of the universe center value, it ensures pressure control accuracy and adapts to high-precision verification requirements. Simultaneously, it ensures that the control quantity remains within the physical limits of the actuator, avoiding equipment damage or control failure, enhancing system robustness. Furthermore, the definability of sets and functions adapts to the nonlinearity and variability of pressure systems, further strengthening the adaptability of fuzzy control.

[0063] Step S400: Obtain the external information of the previous wheel corresponding to the actual pressure of the previous wheel, predict the external information of the current wheel for control based on the external information of the previous wheel, adjust the initial defuzzy control information based on the external information of the current wheel to obtain the target control information, and use the target control information for pressure control.

[0064] The previous round of external information specifically refers to the information presented by the pressure generation system itself and the environmental information it was in during the previous round of pressure control. This information includes, but is not limited to, operating time, usage time, and the condition of the equipment structure. This information can be obtained by using appropriate detection devices during the pressure control process. Since complete external information can only be obtained when the system performs pressure control, a complete picture of the current round's external information is not available each time the system performs pressure control. To achieve more accurate pressure control, the current round's external information needs to be considered. Therefore, the external information for the current round of control can be predicted based on the previous round's information. Specifically, a backpropagation (BP) neural network can be used to process the previous round's external information to predict the current round's external information. The BP neural network is pre-trained with relevant information and can accurately predict external information. This allows for advance knowledge of the overall external information for the current round of pressure control, helping to adjust the initial unfuzzy control information and comprehensively considering external information for more accurate pressure control. The process of adjusting the initial unfuzzy control information based on external information to obtain the target control information includes the following steps: Step S401: Substitute the current external information into the preset external information and pressure influence correspondence model to obtain pressure influence information.

[0065] Step S402: Substitute the pressure influence information into the preset pressure and valve control correspondence to obtain control adjustment information.

[0066] Step S403: Superimpose the control adjustment information onto the initial defuzzified control information to obtain the target control information.

[0067] The preset model for the correspondence between external information and pressure influence, as well as the correspondence between pressure and valve control, are obtained by summarizing historical information. By substituting the current external information into the preset model for the correspondence between external information and pressure influence, pressure influence information is obtained. Then, this pressure influence information is substituting into the preset correspondence between pressure and valve control to obtain control adjustment information. This control adjustment information refers to the influence of external information on pressure control. Finally, superimposing the control adjustment information onto the initial unfuzzy control information yields more accurate target control information. This fuzzy control approach, without requiring a precise mathematical model, can adapt to the nonlinearity and time-varying nature of pressure-generating systems. It also considers external information to further adjust the fuzzy control, achieving precise and stable pressure control and improving the system's robustness and adaptability. Furthermore, after obtaining the target control information, the following steps are also included: Step S500: Obtain the current round actual pressure and the current round preset pressure corresponding to the current round output of the target control information; determine the current round actual pressure error between the current round actual pressure and the current round preset pressure, and the current round actual error change rate corresponding to the current round actual pressure error.

[0068] Step S600: Determine whether the actual pressure error and the rate of change of the actual error in this round meet the standards. If they do, proceed to the next round of pressure control.

[0069] Step S700: If the target is not met, obtain the external information corresponding to the actual pressure of this round, and adjust the output layer weights of the BP neural network based on the external information of this round and the real-time external information to obtain a new deep learning model.

[0070] The actual pressure of the current cycle can be obtained by reading the pressure sensor, and the preset pressure of the current cycle can be obtained by reading the information stored at the control terminal. Then, the difference between the actual pressure of the current cycle and the preset pressure of the current cycle can be obtained, which is the error of the pressure generated by the pressure generation system in the most recent generation.

[0071] Next, determine whether the actual pressure error in this round is less than the preset target pressure error. If so, the actual pressure error in this round meets the target. If not, the actual pressure error in this round does not meet the target.

[0072] Once the actual pressure error in this round is confirmed to be within the acceptable range, it indicates that the pressure control work for this round has been completed, and the next round of pressure control work can then begin.

[0073] If the actual pressure error in this round is determined to be substandard, in order to reduce the impact on the next round of pressure control, the corresponding external information for this round is acquired through a corresponding detection device. Then, based on this external information and real-time external information, the output layer weights of the BP neural network are adjusted to obtain a new deep learning model. This method, by adjusting only the output layer weights of the BP neural network, speeds up the adjustment process, thereby reducing the impact on the next round of pressure control. This adjustment can be completed during steps S100 to S300 to obtain a more accurate deep learning model. Furthermore, this relatively simple method of adjusting the BP neural network can compensate for inaccuracies in the process of determining the initial unfuzzy control information, reducing the need to adjust the models or rule bases involved in determining the initial unfuzzy control information, thus indirectly and quickly obtaining accurate control pressure.

[0074] Figure 4 This is a connection diagram of a pressure control system provided in this application. Figure 4 As shown, a pressure control system includes: a pressure detection module, a pressure processing module, and a pressure control module.

[0075] The pressure detection module is used to acquire the previous round's actual pressure and the corresponding preset pressure, determine the error between the actual and preset pressures, and the rate of change of the actual pressure error. The pressure processing module fuzzifies the previous round's actual pressure error and rate of change to obtain fuzzy upper-round pressure information; it then acquires a fuzzy control rule base, substitutes the fuzzy upper-round pressure information into it to obtain initial fuzzy control information, and defuzzifies the initial fuzzy control information to obtain initial defuzzified control information. The pressure control module acquires the previous round's external information corresponding to the output previous round's actual pressure, predicts the current round's external information based on this information, adjusts the initial defuzzified control information based on this information to obtain target control information, and uses this target control information for pressure control.

[0076] The other functions performed by the pressure detection module, pressure processing module, and pressure control module, as well as the technical details of each function, are the same as or similar to the corresponding features in the basic pressure control method described above, so they will not be repeated here.

[0077] This application also provides a computer storage medium storing a computer program that, when run on a computer, enables the computer to execute the steps in a pressure control method described above.

[0078] It should be understood that although the steps in the flowcharts in the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order requirement for the execution of these steps, and they can be performed in other orders.

[0079] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A pressure control method, characterized in that, The method includes: Obtain the previous round's actual pressure and the corresponding previous round's preset pressure, determine the error between the previous round's actual pressure and the previous round's preset pressure, and the rate of change of the previous round's actual pressure error corresponding to the previous round's actual pressure error; The fuzzy upper wheel pressure information is obtained by fuzzifying the actual pressure error of the upper wheel and the rate of change of the actual error of the upper wheel. Obtain the fuzzy control rule base, substitute the fuzzy upper wheel pressure information into the fuzzy control rule base to obtain the initial fuzzy control information, and perform defuzzification processing on the initial fuzzy control information to obtain the initial defuzzy control information; The system acquires and outputs the external information of the previous wheel corresponding to the actual pressure of the previous wheel, predicts the external information of the current wheel for control based on the external information of the previous wheel, adjusts the initial unfuzzy control information based on the external information of the current wheel to obtain the target control information, and uses the target control information for pressure control.

2. The method according to claim 1, characterized in that, The blurred upper wheel pressure information includes blurred pressure error and blurred error change rate. The process of blurring the actual upper wheel pressure error and the actual upper wheel error change rate to obtain the blurred upper wheel pressure information includes: Obtain the error fuzzy set and the first error membership function corresponding to the pressure error attribute, and the change rate fuzzy set and the first error change rate membership function corresponding to the pressure error change rate attribute; The first error membership function is used to process the actual pressure error of the upper wheel to obtain the first error membership information, wherein the first error membership information includes several first error membership sub-information; Obtain the first reference error membership sub-information representing the high strength of membership in the first error membership information, determine whether there is a strong correlation between the first reference error membership sub-information, if not, substitute the highest strength first reference error membership sub-information in the first error membership information into the error fuzzy set to obtain the fuzzy pressure error; If so, obtain the second error membership function corresponding to the pressure error attribute, use the second error membership function to process the actual pressure error of the previous round to obtain the second error membership information, and determine the fuzzy pressure error based on the first error membership information and the second error membership information; The first error change rate membership function is used to process the actual error change rate of the previous round to obtain the first error change rate membership information, wherein the first error change rate membership information includes several first error change rate membership sub-information. Obtain the first reference error change rate sub-information representing the high intensity of the first error change rate membership information, determine whether there is a strong correlation between the first reference error change rate sub-information, if not, substitute the highest intensity first reference error change rate sub-information in the first error change rate membership information into the change rate fuzzy set to obtain the fuzzy error change rate. If so, obtain the second error change rate membership function corresponding to the pressure error change rate attribute, use the second error change rate membership function to process the previous round's actual error change rate to obtain the second error change rate membership information, and determine the fuzzy error change rate based on the first error change rate membership information and the second error change rate membership information.

3. The method according to claim 2, characterized in that, The second error membership information includes several second error membership sub-information, and the error fuzzy set includes several error fuzzy languages, each error membership sub-information corresponding to one error fuzzy language. Determining the fuzzy pressure error based on the first and second error membership information includes: Obtain the first weight corresponding to the first error membership function, and multiply the first weight by each first error membership sub-information to obtain the first discrimination value corresponding to each first error membership sub-information; Obtain the second weight corresponding to the second error membership function, and multiply the second weight by each second error membership sub-information to obtain the second discrimination value corresponding to each second error membership sub-information; The first and second distinction values ​​corresponding to the same error fuzzy language are added together to obtain the distinction value of the error fuzzy language, and the error fuzzy language corresponding to the largest distinction value among all distinction values ​​is determined as the fuzzy pressure error.

4. The method according to claim 1, characterized in that, The fuzzy control rule base includes: Acquire historical fuzzy pressure information and corresponding historical fuzzy control information that have met the standards in the completed pressure control work; The corresponding historical fuzzy pressure information and historical fuzzy control information are divided into training information group, verification information group and test information group; The initial neural network is trained using the training information set to obtain the model to be validated. The verification information group is used to verify the model to be verified, and the model to be verified is adjusted to obtain the model to be used. The test information group is used to test the model to be used to obtain test results. The test results are then used to determine whether the test results are qualified. If they are qualified, the model to be used is determined as a fuzzy control rule library. If it fails, the corresponding historical fuzzy pressure information and historical fuzzy control information are reclassified into training information group, verification information group and test information group.

5. The method according to claim 1, characterized in that, The process of defuzzifying the initial fuzzy control information to obtain the initial defuzzified control information includes: Obtain the set of fuzzy control quantities and the membership function of the control quantities corresponding to the initial fuzzy control information, wherein the fuzzy set of control quantities includes several fuzzy linguistic terms that characterize different control intensities, and each fuzzy linguistic term corresponds to a membership distribution interval in the membership function of the control quantities; Obtain the membership degree value corresponding to each fuzzy linguistic item in the initial fuzzy control information, and establish the relationship between the fuzzy linguistic item and the membership degree value based on the membership degree value; Based on the membership function of the control quantity, the universe center value corresponding to each fuzzy linguistic item is determined, and the membership value and the universe center value are processed by the weighted average method to obtain the initial solution fuzzy control information.

6. The method according to claim 1, characterized in that, The method of predicting the external information of the current round of control based on the external information of the previous round includes: using a BP neural network to process the external information of the previous round to predict the external information of the current round of control.

7. The method according to claim 1, characterized in that, The step of adjusting the initial unfuzzy control information based on the current external information to obtain the target control information includes: Substitute the current external information into a preset model corresponding to external information and pressure effects to obtain pressure effect information; The pressure influence information is substituted into a preset pressure-valve control correspondence to obtain control adjustment information; The target control information is obtained by superimposing the control adjustment information onto the initial unfuzzy control information.

8. The method according to claim 6, characterized in that, After obtaining the target control information, the process also includes: Obtain the actual pressure of the current round output corresponding to the target control information and the preset pressure of the current round corresponding to the actual pressure of the current round, and determine the actual pressure error of the current round between the actual pressure of the current round and the preset pressure of the current round; Determine whether the actual pressure error of this round meets the standard. If it does, proceed to the next round of pressure control. If the target is not met, obtain the external information corresponding to the actual pressure of this round, and adjust the output layer weights of the BP neural network based on the external information of this round and the real-time external information to obtain a new deep learning model.

9. A pressure control system, characterized in that, The system includes: a pressure detection module, a pressure processing module, and a pressure control module; wherein... The pressure detection module is used to obtain the previous wheel actual pressure output and the previous wheel preset pressure corresponding to the previous wheel actual pressure, and to determine the previous wheel actual pressure error between the previous wheel actual pressure and the previous wheel preset pressure and the previous wheel actual error change rate corresponding to the previous wheel actual pressure error. The pressure processing module is used to perform fuzzification processing on the actual pressure error of the upper wheel and the change rate of the actual error of the upper wheel to obtain fuzzy upper wheel pressure information; obtain a fuzzy control rule base, substitute the fuzzy upper wheel pressure information into the fuzzy control rule base to obtain initial fuzzy control information, and perform defuzzification processing on the initial fuzzy control information to obtain initial defuzzy control information. The pressure control module is used to acquire the external information of the previous wheel corresponding to the actual pressure of the previous wheel, predict the external information of the current wheel based on the external information of the previous wheel, adjust the initial defuzzification control information based on the external information of the current wheel to obtain target control information, and use the target control information to perform pressure control.

10. A computer-readable storage medium having a computer program stored thereon that can run on a processor, characterized in that, When the computer program is executed by the processor, it implements a pressure control method as described in any one of claims 1 to 8.

Images