Delivery pump pressure estimation method and system based on dynamic multi-objective optimization KF

Abstract

The invention discloses a dynamic multi-objective optimization KF-based delivery pump pressure estimation method and system, and the method comprises the steps: constructing a discrete state space equation and a measurement equation of the pressure of a stock solution delivery pump in a carbon fiber polymerization process; calculating a pressure priori estimation value and an error covariance matrix priori estimation value of the stock solution delivery pump in the carbon fiber polymerization process; calculating a Kalman gain, a delivery pump pressure posteriori estimation value and an error covariance matrix posteriori estimation value; then, a dynamic multi-objective optimization algorithm is adopted to solve the defined multi-objective function, and a Kalman filtering process noise covariance and a measurement noise covariance in a dynamic environment are rapidly optimized; and then, selecting an optimal decision solution in the Pareto non-dominated solution set, constructing an optimal noise matrix covariance matrix and a measurement noise covariance matrix in the Kalman filtering process, and applying the optimal noise matrix covariance matrix and the measurement noise covariance matrix to the pressure estimation of the stock solution delivery pump to realize the real-time accurate estimation of the pressure of the stock solution delivery pump in the carbon fiber polymerization process.

Description

Technical Field

[0001] The present application relates to the technical field of carbon fiber production process control, and in particular to a method and system for estimating the pressure of a raw liquid delivery pump in a carbon fiber polymerization process based on dynamic multi-objective optimization Kalman filtering (KF). Background Art

[0002] During the carbon fiber polymerization process, raw materials such as acrylonitrile and initiator undergo polymerization reaction, desinging, and degassing to generate spinning dope. After that, they need to be transported to the subsequent carbon fiber spinning process in a stable and accurate manner through a dope delivery pump, and then undergo coagulation, forming, and stretching to form primary fibers.

[0003] Estimating the pressure of the dope pump is crucial for ensuring smooth delivery of the spinning dope and the uniformity and consistency of the resulting spun fiber yarn. As a key parameter characterizing the dope pump's operational performance, process stability, and the quality of the resulting spun fiber, accurate estimation not only effectively improves on-site production safety but also significantly reduces energy consumption and wear on the pump equipment, thereby ensuring uniformity and consistency in the resulting spun fiber yarn.

[0004] However, the existing technology relies on pressure data measured by sensors built into or installed on-site in the delivery pump. During the actual delivery of spinning dope by the dope delivery pump, the stability of its pressure parameters is easily disturbed by factors such as changes in production conditions, external environmental noise, and aging of delivery pump components, resulting in an increase in the pressure estimation error of the dope delivery pump during the carbon fiber polymerization process, causing the flow rate of the spinning dope to fluctuate or deviate violently, thereby affecting the quality of the nascent fiber product.

[0005] Although some liquid conveying pipeline structures that improve flow stability can meet the needs of conveying operations to a certain extent, their structures are complex and fixed, and they cannot be adjusted in real time according to different material conveying requirements, and their flexibility and adaptability are insufficient.

[0006] Some delivery pump outlet pressure stabilization compensation devices need to consume a certain amount of energy to maintain the balance of delivery pump outlet pressure, but this increases the company's energy consumption and operating costs. Summary of the Invention

[0007] The purpose of this application is to provide a method and system for estimating the pressure of a delivery pump based on dynamic multi-objective optimization KF, so as to improve the accuracy and real-time performance of the pressure estimation of the raw liquid delivery pump in the carbon fiber polymerization process.

[0008] On this basis, it has high flexibility, scalability and economy, and can better cope with dynamic and complex working conditions on site.

[0009] In order to solve the above problems, the present invention adopts the following technical solutions:

[0010] A method for estimating the pressure of a delivery pump based on dynamic multi-objective optimization KF, characterized in that it is used for the raw liquid delivery in the carbon fiber polymerization process, and includes the following steps:

[0011] S1: Collect the historical pressure data of the delivery pump, construct the discrete state space equation and measurement equation of the raw liquid delivery pump pressure in the carbon fiber polymerization process, and set the initial values of the system process noise covariance matrix, measurement noise covariance matrix and error covariance matrix for the initial state update and operation calculation of the system;

[0012] S2: Calculate and obtain the prior estimate value of the delivery pump pressure and the prior estimate value of the error covariance matrix at the current moment through the Kalman filter prediction equation;

[0013] S3: Calculate and obtain the Kalman gain, the posterior estimate value of the delivery pump pressure and the posterior estimate value of the error covariance matrix at the current moment through the Kalman filter correction equation;

[0014] S4: On the basis of obtaining the posterior estimate value of the delivery pump pressure, use the dynamic multi-objective optimization algorithm to solve the defined multi-objective function, and quickly optimize the process noise covariance and measurement noise covariance in the Kalman filter process in the dynamic environment;

[0015] S5: Adopt the Pareto optimal decision-making mechanism based on the ideal point, select the optimal decision solution in the Pareto non-dominated solution set, and use it to construct the optimal noise matrix covariance matrix and measurement noise covariance matrix in the Kalman filter process, and apply it to the carbon fiber polymerization process at the current moment for the pressure estimation of the raw liquid delivery pump;

[0016] S6: At the next moment, repeat the above Kalman filter, dynamic multi-objective algorithm fast optimization and decision-making process to realize the real-time and accurate estimation of the pressure of the raw liquid delivery pump in the carbon fiber polymerization process.

[0017] In the above technical solution, the construction of the discrete state space equation and the measurement equation in step S1 includes defining the state vector, state matrix, input matrix, input vector, system observation matrix, and setting the corresponding initial values.

[0018] In the above technical solution, the discrete state space equation of the raw liquid delivery pump pressure in the carbon fiber polymerization process in step S1 is configured as:

[0019] The pressure state vector of the raw liquid delivery pump in the carbon fiber polymerization process at the current moment is equal to the sum of the following three parts: the value of the pressure state vector at the previous moment multiplied by the state matrix, the pressure input vector at the previous moment multiplied by the input matrix, and the process noise of the system at the previous moment.

[0020] In the above technical solution, the measurement equation in step S1 is configured as: the measured value of the pressure sensor of the carbon fiber polymerization process stock solution transfer pump at the current moment is equal to the system observation matrix multiplied by the state vector at the current moment, plus the measurement noise at the current moment.

[0021] In the above technical solution, the Kalman filter prediction equation and correction equation in step S2 include calculating the prior estimate value, the prior estimate value of the error covariance matrix, the Kalman gain, the posterior estimate value, and the posterior estimate value of the error covariance matrix.

[0022] In the above technical solution, step S2 includes: multiplying the pressure input vector of the carbon fiber polymerization process stock solution transfer pump at the previous moment by the input matrix, and adding the posterior estimate value of the transfer pump pressure at the previous moment multiplied by the state matrix, to obtain the prior estimate value of the transfer pump pressure at the current moment predicted at the previous moment.

[0023] In the above technical solution, step S2 includes: converting the posterior estimate value of the error covariance matrix at the previous moment through the state transition matrix A, and adding the process noise covariance matrix at the current moment, to obtain the prior estimate value of the updated prediction error covariance matrix at the current moment.

[0024] In the above technical solution, step S3 includes:

[0025] First, determine the Kalman gain, which is calculated by multiplying the prior error covariance matrix by the transpose of the system observation matrix, and then dividing by the sum of the result of multiplying the system observation matrix by the prior error covariance matrix and the measurement noise covariance matrix;

[0026] Use the Kalman gain calculated in the previous step to update the state estimate: add the previous state estimate to the product of the Kalman gain and the difference between the new measurement value and the previous state estimate converted by the system observation matrix;

[0027] Multiply the identity matrix minus the product of the Kalman gain and the system observation matrix by the prior error covariance matrix to complete the update of the error covariance matrix.

[0028] In the above technical solution, step S4 includes:

[0029] Limit the specific values of the process noise covariance and the measurement noise covariance within a set range to construct decision variables;

[0030] Construct a multi-objective function, including: the first objective function is expressed as the error value between the optimal filtered output value and the true value of the pressure of the carbon fiber polymerization process stock solution transfer pump at the current k moment; the second objective function is expressed as the maximum absolute value of the optimal filtered output value of the pressure of the carbon fiber polymerization process stock solution transfer pump at the current sampling moment k and within a previous time window.

[0031] Use the population-based NSGA-II algorithm to solve, and obtain a set of Pareto non-dominated solutions. Each solution represents the best balance point found between two objective functions.

[0032] In the above technical solution, in step S5, the ideal point includes the minimum values of the two objective functions. By comparing the distance between each solution and the ideal point, the difference between each solution and the minimum values of the solutions in the Pareto non-dominated solution set on the two objective functions is weighted to select the optimal decision solution.

[0033] A carbon fiber polymerization transfer pump pressure estimation system based on dynamic multi-objective optimization KF, characterized by comprising:

[0034] A data acquisition module that acquires the historical pressure data of the transfer pump;

[0035] Usually, a pressure sensor for real-time and accurate monitoring of the change in the pressure of the stock solution is built in at the outlet of the stock solution transfer pump during the carbon fiber polymerization process. The pressure sensor transmits the monitored pressure value to the data acquisition module installed in the industrial control computer through the field industrial Ethernet for storage at high speed;

[0036] A state space equation construction module that constructs the discrete state space equation and measurement equation of the pressure of the stock solution transfer pump during the carbon fiber polymerization process based on historical data, and sets the initial values of the system process noise covariance matrix, measurement noise covariance matrix, and error covariance matrix;

[0037] A Kalman filter module that is used to calculate the prior estimate value of the transfer pump pressure and the prior estimate value of the error covariance matrix through the Kalman filter prediction equation; calculate the Kalman gain, the posterior estimate value of the transfer pump pressure, and the posterior estimate value of the error covariance matrix through the Kalman filter correction equation;

[0038] A dynamic multi-objective optimization module that uses a dynamic multi-objective optimization algorithm to solve the defined multi-objective function and quickly optimize the process noise covariance and measurement noise covariance of the Kalman filter in a dynamic environment;

[0039] A Pareto optimal decision module that uses a Pareto optimal decision-making mechanism based on the ideal point to select the optimal decision solution in the Pareto non-dominated solution set, and is used to construct the optimal noise matrix covariance matrix and measurement noise covariance matrix of the Kalman filter and apply them to the pressure estimation of the stock solution transfer pump during the carbon fiber polymerization process at the current moment.

[0040] A computer-readable storage medium, characterized in that a computer program for implementing the method described in any one of the above is stored thereon.

[0041] In summary, the present invention aims to solve the problem that traditional pressure estimation methods for transfer pumps are difficult to effectively cope with the interference of external environmental noise and the aging of transfer pump components under complex dynamic working conditions, resulting in large errors in the pressure estimation of transfer pumps. A pressure estimation method for the transfer pump in the polymerization process based on dynamic multi-objective optimization KF is provided. By making full use of the fast search characteristics of the dynamic multi-objective optimization algorithm, the process noise covariance and measurement noise covariance of the Kalman filter are dynamically optimized. The Pareto optimal decision-making mechanism based on the ideal point is adopted to select the optimal noise matrix covariance matrix and measurement noise covariance matrix for constructing the Kalman filter, and they are applied to the pressure estimation of the raw liquid transfer pump at the current sampling moment to achieve real-time and accurate estimation of the pressure of the raw liquid transfer pump in the carbon fiber polymerization process.

[0042] Compared with the prior art, the present application has at least the following beneficial effects:

[0043] 1) By making full use of the fast search characteristics of the dynamic multi-objective optimization algorithm, it can quickly and evenly cover the entire Pareto front in the time-varying objective space, realize the dynamic optimization of the process noise covariance and measurement noise covariance of the Kalman filter, and effectively improve the accuracy and real-time performance of the pressure estimation of the raw liquid transfer pump in the carbon fiber polymerization process. Even under complex conditions such as changes in production working conditions, external environmental noise, and the aging of transfer pump components, this method can estimate the pressure of the transfer pump more accurately.

[0044] 2) The Pareto optimal decision-making mechanism based on the ideal point is adopted to select a point closest to the ideal point in the objective space from the Pareto non-dominated solution set as the decision solution, which is used to construct the optimal noise matrix covariance matrix and measurement noise covariance matrix of the Kalman filter, ensuring the best global balance between the two defined objective functions; at the same time, by adjusting the Kalman filter parameters in real time, external interference can be effectively resisted, making the pressure estimation more stable, thereby enhancing the adaptability of the system to abnormal situations and significantly enhancing the adaptability and robustness of the pressure estimation method for the raw liquid transfer pump in the carbon fiber polymerization process.

[0045] 3) Compared with the method that relies on purchasing additional equipment to estimate the pressure of the transfer pump, the pressure estimation method for the transfer pump based on dynamic multi-objective optimization KF does not require additional equipment support, such as pressure stabilizing valves, nitrogen buffer tanks, etc., reducing the equipment purchase and maintenance costs, and having high flexibility, scalability, and economy, and being able to better cope with the on-site dynamic and complex working conditions.

[0046] 4) This method does not rely on a fixed structure of a certain polymerization device, has high flexibility and scalability, and can adapt to different production requirements and working conditions changes. It has high flexibility, scalability and economy, and can better cope with the dynamic and complex on-site working conditions. Through stable pressure control, it realizes the real-time and accurate estimation of the pressure of the transfer pump, providing the possibility for the real-time monitoring and adjustment of the production process. It can reduce the fluctuation of the spinning dope flow rate, contribute to improving the uniformity and consistency of the as-spun fiber finished filament, and thus ensure the quality of the final product. Brief Description of the Drawings

[0047] To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0048] Figure 1 It is a flow chart of the method for estimating the pressure of the transfer pump in the polymerization process based on dynamic multi-objective optimization KF of the present invention.

[0049] Figure 2 It is an effect diagram of the method for estimating the pressure of the transfer pump in the polymerization process based on dynamic multi-objective optimization KF of the present invention. Detailed Embodiments

[0050] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Usually, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0051] Therefore, the detailed description of the embodiments of the present application provided in the following drawings is not intended to limit the scope of the present application to be protected, but only represents the selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0052] It should be noted that: similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0053] In the description of the present application, it should be noted that the orientation or positional relationship indicated by terms such as "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of this application is habitually placed during use. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present application. In addition, terms such as "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0054] The features and performance of the present application will be further described in detail below in conjunction with embodiments.

[0055] Embodiment 1

[0056] As Figure 1 shown, the method for estimating the pressure of the polymerization process transfer pump based on dynamic multi-objective optimization KF in the embodiment of the present invention includes the following steps:

[0057] Step 1: On the basis of fully analyzing the research object of the pressure of the raw liquid transfer pump in the carbon fiber polymerization process, construct the discrete state space equation and measurement equation of the pressure of the raw liquid transfer pump in the carbon fiber polymerization process, and set the initial value of the system process noise covariance matrix, the initial value of the measurement noise covariance matrix, and the initial value of the error covariance matrix;

[0058] Step 2: At the current sampling moment, calculate the prior estimate value of the pressure of the raw liquid transfer pump in the carbon fiber polymerization process and the prior estimate value of the error covariance matrix through the Kalman Filter (KF) prediction equation;

[0059] Step 3: Calculate the Kalman gain, the posterior estimate value of the transfer pump pressure, and the posterior estimate value of the error covariance matrix through the Kalman filter correction equation;

[0060] Step 4: On the basis of obtaining the posterior estimate value of the transfer pump pressure, use the dynamic multi-objective optimization algorithm to solve the defined multi-objective function, and quickly optimize the process noise covariance and measurement noise covariance in the Kalman filtering process in a dynamic environment;

[0061] Step 5: Adopt the Pareto optimal decision-making mechanism based on the ideal point, select the optimal decision solution in the Pareto non-dominated solution set, and use it to construct the optimal noise matrix covariance matrix and measurement noise covariance matrix in the Kalman filtering process, and apply them to the pressure estimation of the raw liquid transfer pump in the carbon fiber polymerization process at the current sampling moment;

[0062] Step 6: At the next sampling moment, repeat the above Kalman filtering, dynamic multi-objective algorithm fast optimization, and decision-making process to achieve real-time and accurate estimation of the pressure of the stock solution delivery pump in the carbon fiber polymerization process.

[0063] Furthermore, in Step 1, constructing the discrete state space equation and measurement equation of the pressure of the stock solution delivery pump in the carbon fiber polymerization process, and setting the initial values of the system process noise covariance matrix, measurement noise covariance matrix, and error covariance matrix include the following steps:

[0064] Step 1-1: Construct the discrete state space equation and measurement equation of the pressure of the stock solution delivery pump in the carbon fiber polymerization process. Among them, the discrete state space equation is defined as follows:

[0065] x k = Ax k-1 + Bu k-1 + w k-1 ;

[0066] In the formula, x k is a 1×1 dimensional state vector representing the pressure of the stock solution delivery pump in the carbon fiber polymerization process; A represents a 1×1 dimensional state matrix; x k-1 represents the state vector value at the (k - 1) sampling moment; B represents a 1×1 dimensional input matrix; u k-1 represents a 1×1 dimensional input vector; w k-1 represents the process noise of the system at the (k - 1) sampling moment. In the present invention, w k-1 obeys a normal distribution with an expectation of 0 and a process noise covariance matrix of Q k-1 , that is, w k-1 ~ N(0, Q k-1 ).

[0067] Based on a full analysis of the research object of the pressure of the stock solution delivery pump in the carbon fiber polymerization process, the discrete state space equation can be further defined as follows:

[0068]

[0069] In the formula, ε represents the mutation threshold factor; Δx k represents the mutation amplitude. At this time, the state matrix A = 1, and the input matrix B = 1. At the k moment, when x k - x k-1 ≤ ε, the pressure of the delivery pump in the polymerization process is in a stable state, and x k does not mutate; when x k - x k-1 > ε, the pressure of the delivery pump in the polymerization process is in a mutation state, and x k mutates, and its value will increase by Δx k .

[0070] The measurement equation is defined as follows:

[0071] z k = H m x k + v k ;

[0072] where z k represents the sensor measurement value of 1×1 dimension at time k; H m represents the system observation matrix of 1×1 dimension. Since the value z k measured by the sensor represents the pressure value of the delivery pump, so H m = 1; x k represents the state vector at time k; v k represents the measurement noise at time k. In the present invention, v k obeys a normal distribution with an expectation of 0 and a measurement noise covariance matrix of R k , that is, v k ~ N(0, R k ).

[0073] Step 1-2: Set the initial value of the system process noise covariance matrix Q0 = [1], the initial value of the measurement noise covariance matrix R0 = [1], and the initial value of the error covariance matrix P0 = [1] for the initial state update and operation calculation of the system at the sampling time k = 1.

[0074] Furthermore, in Step 2, calculating the prior estimate value of the pressure of the carbon fiber polymerization process stock solution delivery pump and the prior estimate value of the error covariance matrix includes the following steps:

[0075] Step 2-1: At the current sampling time, calculate the prior estimate value of the pressure of the carbon fiber polymerization process stock solution delivery pump through the Kalman filter prediction equation The calculation formula for the prior estimate value of the delivery pump pressure is as follows:

[0076]

[0077] where represents the prior estimate value of the delivery pump pressure at time k predicted at time k - 1; A represents the state matrix of 1×1 dimension; represents the posterior estimate value of the delivery pump pressure at time k - 1; B represents the input matrix of 1×1 dimension; u k-1 represents the input vector of 1×1 dimension;

[0078] Step 2-2: At the current system sampling time, calculate the prior estimate value of the error covariance matrix through the Kalman filter prediction equation. The calculation formula for the prior estimate value of the error covariance matrix is as follows:

[0079]

[0080] In the formula, represents the prior estimated value of the error covariance matrix at time k; A represents the state matrix of 1×1 dimension; P k-1 represents the posterior estimated value of the error covariance matrix at time k-1; Q k represents the process noise covariance matrix at time k.

[0081] Furthermore, in step 3, calculating the Kalman gain, the posterior estimated value of the delivery pump pressure, and the posterior estimated value of the error covariance matrix includes the following steps:

[0082] Step 3-1: Calculate the Kalman gain through the Kalman filter correction equation. The calculation formula of the Kalman gain is as follows:

[0083]

[0084] In the formula, K k represents the Kalman gain at time k; represents the prior estimated value of the error covariance matrix at time k; H m represents the system observation matrix of 1×1 dimension; R k represents the measurement noise covariance matrix at time k;

[0085] Step 3-2: Calculate the posterior estimated value of the delivery pump pressure through the Kalman filter correction equation. The calculation formula of the posterior estimated value of the delivery pump pressure is as follows:

[0086]

[0087] In the formula, represents the optimal filtered output value of the delivery pump pressure calculated after combining the Kalman filter with the discrete state space equation and the measured value at time k; represents the prior estimated value of the delivery pump pressure at time k predicted at time k-1; K k represents the Kalman gain at time k; z k represents the sensor measurement value of 1×1 dimension at time k; H m represents the system observation matrix of 1×1 dimension;

[0088] Step 3-3: Calculate the posterior estimated value of the error covariance matrix through the Kalman filter correction equation. The calculation formula of the posterior estimated value of the error covariance matrix is as follows:

[0089]

[0090] In the formula, P k represents the posterior estimated value of the error covariance matrix updated at time k; I represents the identity matrix; Kk represents the Kalman gain at time k; H m represents the system observation matrix of 1×1 dimension; P k - represents the prior estimate of the error covariance matrix at time k.

[0091] Furthermore, in step 4, solving the defined multi-objective function using the dynamic multi-objective optimization algorithm includes the following steps:

[0092] Step 4-1, construct the decision variables in the dynamic multi-objective optimization algorithm. Since the discrete state space equation only involves one-dimensional cases, both the process noise covariance matrix Q k and the measurement noise covariance matrix R k have only one element, that is, Q k = diag(q k ) and R k = diag(r k ). Therefore, define the two-dimensional decision variable as [q k , r k , and their respective numerical ranges are as follows:

[0093] q min ≤ q k ≤ q max ;

[0094] r min ≤ r k ≤ r max ;

[0095] In the formula, q max , q min respectively represent the upper and lower limit values of q k ; r max , r min respectively represent the upper and lower limit values of r k . The two-dimensional decision variable defined above will be encoded into the chromosomes of individuals in the population of the dynamic multi-objective optimization algorithm, and then constitute the candidate solutions in the optimization problem space;

[0096] Step 4-2, in order to optimize q k and r kEvaluate the candidate solution set composed of different combinations to construct the multi-objective function in the dynamic multi-objective optimization algorithm. In the present invention, two objective functions are constructed. Among them, the first objective function is expressed as the error value between the optimal filtered output value and the true value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process at the current k moment, so as to measure the real-time performance of the pressure estimation method and guide the algorithm to quickly optimize the process noise covariance matrix and measurement noise covariance matrix in the Kalman filtering process in a dynamic environment. Considering that due to the interference of external environmental noise, the accurate true value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process cannot be obtained in actual production on site, the average value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process within a period of time window is used in the present invention to approximately replace the true value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process. The first objective function is defined as follows:

[0097]

[0098] In the formula, q k represents the element in the process noise covariance matrix Q k ; r k represents the element in the measurement noise covariance matrix R k ; k represents the current sampling moment; represents the optimal filtered output value of the transfer pump pressure calculated after the Kalman filter combines the discrete state space equation estimate value and the measurement value at the k moment; N1 represents a period of time window; f1 represents a mapping function about the parameters q k , r k and the current sampling moment k.

[0099] Under the complex dynamic working conditions on site, the pressure of the stock solution transfer pump in the carbon fiber polymerization process is often affected by factors such as external environmental noise and aging of the transfer pump components, resulting in drastic fluctuations. Therefore, in order to make the estimated value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process change smoothly, the second objective function is expressed as the maximum absolute value of the optimal filtered output value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process at the current sampling moment k and within the previous period of time window, guiding the algorithm to search for the solution area with the smallest fluctuation of the estimated value, ensuring the adaptability and robustness of the pressure estimation method. The second objective function is defined as follows:

[0100]

[0101] In the formula, k represents the current sampling moment; N2 represents a period of time window, f2 represents a mapping function about the parameters q k , r k and the current sampling moment k;

[0102] Step 4-3: Based on the decision variable encoding and the definition of the objective function, the population-based dynamic multi-objective optimization algorithm NSGA-II (Non-dominated Sorting Genetic Algorithm II) is used to solve the defined multi-objective function. This algorithm optimizes the decision variables through operations such as crossover and mutation, guides the population search in the algorithm for the Pareto front, and at the same time combines evolutionary strategies such as non-dominated sorting and crowding distance to maintain the distribution and uniformity of the solutions, so as to achieve the rapid solution of the multi-objective optimization problem. After using the dynamic multi-objective optimization algorithm to solve the multi-objective function, a set of Pareto non-dominated solution sets for the decision variables [q k , r k will be obtained. Each solution in this solution set represents the trade-off relationship between the two defined objective functions.

[0103] Furthermore, in step 5, a Pareto optimal decision-making mechanism based on the ideal point is adopted to select the optimal decision solution from the Pareto non-dominated solution set, which is used to construct the optimal noise matrix covariance matrix and the measurement noise covariance matrix in the Kalman filtering process and is applied to the pressure estimation of the raw liquid transfer pump at the current sampling moment, including the following steps:

[0104] Step 5-1: The formula for the Pareto optimal decision-making mechanism based on the ideal point to select the optimal decision solution from the Pareto non-dominated solution set is defined as follows:

[0105]

[0106] In the formula, represents the optimal decision solution selected from the Pareto non-dominated solution set at the current sampling moment k; I represents the Pareto non-dominated solution set; i represents a solution in the Pareto non-dominated solution set; represents the minimum value of all solutions in the Pareto non-dominated solution set with respect to the first defined objective function; represents the minimum value of all solutions in the Pareto non-dominated solution set with respect to the second defined objective function; i1 represents the value of the solution i in the Pareto non-dominated solution set on the first defined objective function; i2 represents the value of the solution i in the Pareto non-dominated solution set on the second defined objective function; α1 represents the preference factor on the first defined objective function; α2 represents the preference factor on the second defined objective function; F1 represents the difference between the maximum value and the minimum value of all solutions in the Pareto non-dominated solution set with respect to the first defined objective function; F2 represents the difference between the maximum value and the minimum value of all solutions in the Pareto non-dominated solution set with respect to the second defined objective function;

[0107] Step 5-2: Select the optimal decision solution from the Pareto non-dominated solution set at the current sampling time k For constructing the optimal process noise matrix covariance matrix of the Kalman filter And the measurement noise covariance matrix And apply it to the pressure estimation of the carbon fiber polymerization process stock solution transfer pump at the current sampling time.

[0108] Embodiment 2

[0109] Figure 2 Shows the effect diagram of the pressure estimation method of the transfer pump in the polymerization process based on dynamic multi-objective optimization KF. The abscissa is time and the ordinate is pressure. It can be seen that the original transfer pump pressure has not undergone any filtering and optimization processing, and the pressure signal has relatively large noise. If the traditional estimation method that relies on the measurement data of the pressure sensor is directly used, the pressure estimation error of the transfer pump is relatively large.

[0110] However, by introducing the dynamic multi-objective optimization algorithm and the Kalman filter method based on the ideal point Pareto optimal decision-making mechanism, through the optimization and decision-making of the process noise covariance and the measurement noise covariance, the interference signal can be fully filtered, and the filtered pressure data is relatively smooth, effectively improving the accuracy and real-time performance of the pressure estimation of the carbon fiber polymerization process stock solution transfer pump.

[0111] The embodiments described above are some, but not all, of the embodiments of the present application. The detailed description of the embodiments of the present application is not intended to limit the scope of the present application claimed, but merely represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

Claims

1. A pressure estimation method for a delivery pump based on dynamic multi-objective optimization KF, characterized in that, For the raw liquid transportation in the carbon fiber polymerization process, it includes the following steps: S1: Collect the historical pressure data of the transfer pump, construct the discrete state space equation and measurement equation of the raw liquid transfer pump pressure in the carbon fiber polymerization process, and set the initial values of the system process noise covariance matrix, measurement noise covariance matrix, and error covariance matrix for the initial state update and operation calculation of the system; S2: Calculate and obtain the prior estimated value of the transfer pump pressure and the prior estimated value of the error covariance matrix at the current moment through the Kalman filter prediction equation; S3: Calculate and obtain the Kalman gain, the posterior estimated value of the transfer pump pressure, and the posterior estimated value of the error covariance matrix at the current moment through the Kalman filter correction equation; S4: Based on obtaining the posterior estimated value of the transfer pump pressure, use the dynamic multi-objective optimization algorithm to solve the defined multi-objective function, and quickly optimize the process noise covariance and measurement noise covariance in the Kalman filter under the dynamic environment; S5: Adopt the Pareto optimal decision-making mechanism based on the ideal point, select the optimal decision solution in the Pareto non-dominated solution set, and use it to construct the optimal noise matrix covariance matrix and measurement noise covariance matrix in the Kalman filter process, and apply them to the carbon fiber polymerization process at the current moment for the pressure estimation of the raw liquid transfer pump; S6: At the next moment, repeat the above Kalman filter, dynamic multi-objective algorithm for quick optimization and decision-making process to achieve real-time and accurate estimation of the raw liquid transfer pump pressure in the carbon fiber polymerization process.

2. The method for estimating the pressure of a delivery pump based on dynamic multi-objective optimization KF according to claim 1, characterized in that, The construction of the discrete state space equation and measurement equation described in step S1 includes defining the state vector, state matrix, input matrix, input vector, system observation matrix, and setting the corresponding initial values.

3. The method for estimating the pressure of a delivery pump based on dynamic multi-objective optimization KF according to claim 1, wherein The Kalman filter prediction equation and correction equation described in step S2 include calculating the prior estimated value, the prior estimated value of the error covariance matrix, the Kalman gain, the posterior estimated value, and the posterior estimated value of the error covariance matrix.

4. The method for estimating the pressure of a delivery pump based on dynamic multi-objective optimization KF according to claim 1, wherein, Step S2 includes: Multiply the input vector of the raw liquid transfer pump pressure in the carbon fiber polymerization process at the previous moment by the input matrix, and add the product of the posterior estimated value of the transfer pump pressure at the previous moment and the state matrix to obtain the prior estimated value of the transfer pump pressure at the current moment predicted at the previous moment.

5. The pressure estimation method of the delivery pump based on dynamic multi-objective optimization KF according to claim 1, characterized in that Step S2 includes: Convert the posterior estimated value of the error covariance matrix at the previous moment through the state transition matrix, and add the process noise covariance matrix at the current moment to obtain the updated prior estimated value of the prediction error covariance matrix at the current moment.

6. The pressure estimation method of the delivery pump based on dynamic multi-objective optimization KF according to claim 1, characterized in that, Step S3 includes: First determine the Kalman gain, which is calculated by multiplying the prior error covariance matrix by the transpose of the system observation matrix, and then dividing by the sum of the result of multiplying the system observation matrix by the prior error covariance matrix and the measurement noise covariance matrix; Use the Kalman gain calculated in the previous step to update the state estimate: Add the previous state estimate to the product of the Kalman gain and the difference between the new measurement value and the previous state estimate converted through the system observation matrix; Complete the update of the error covariance matrix by multiplying the product of the identity matrix minus the product of the Kalman gain and the system observation matrix by the prior error covariance matrix.

7. The pressure estimation method of the delivery pump based on dynamic multi-objective optimization KF according to claim 1, characterized in that Step S4 includes: Construct decision variables by restricting the specific values of the process noise covariance and the measurement noise covariance within a set range. Construct a multi-objective function, including: the first objective function is expressed as the error value between the optimal filtered output value and the true value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process at the current time k; the second objective function is expressed as the maximum absolute value of the optimal filtered output value of the pressure of the stock solution transfer pump in the carbon fiber polymerization process at the current sampling time k and within a previous time window. Use the population-based NSGA-II algorithm to solve and obtain a set of Pareto non-dominated solutions, each of which represents the best balance point found between the two objective functions.

8. The method for estimating the pressure of a delivery pump based on dynamic multi-objective optimization KF according to claim 1, characterized in that In step S5, the ideal point includes the minimum values of the two objective functions. By comparing the distance between each solution and the ideal point, the optimal decision solution is selected by weighting the difference between each solution and the minimum values of the solutions in the Pareto non-dominated solution set on the two objective functions.

9. A carbon fiber polymerization transfer pump pressure estimation system based on dynamic multi-objective optimization KF, characterized in that, Include: A data acquisition module that acquires historical pressure data of the transfer pump. A state space equation construction module that constructs a discrete state space equation and a measurement equation for the pressure of the stock solution transfer pump in the carbon fiber polymerization process based on historical data, and sets the initial values of the system process noise covariance matrix, the measurement noise covariance matrix, and the error covariance matrix. A Kalman filter module that is used to calculate the prior estimate value of the transfer pump pressure and the prior estimate value of the error covariance matrix through the Kalman filter prediction equation; calculate the Kalman gain, the posterior estimate value of the transfer pump pressure, and the posterior estimate value of the error covariance matrix through the Kalman filter correction equation. A dynamic multi-objective optimization module that uses a dynamic multi-objective optimization algorithm to solve the defined multi-objective function and quickly optimize the process noise covariance and the measurement noise covariance in the Kalman filter process in a dynamic environment. A Pareto optimal decision module that uses a Pareto optimal decision-making mechanism based on the ideal point to select the optimal decision solution in the Pareto non-dominated solution set, which is used to construct the optimal noise matrix covariance matrix and the measurement noise covariance matrix in the Kalman filter process and is applied to the pressure estimation of the stock solution transfer pump in the carbon fiber polymerization process at the current time.

10. A computer-readable storage medium, characterized in that, A computer program for implementing the method according to any one of claims 1-8 is stored thereon.

Images