High-precision mass flow controller adjusting method based on artificial intelligence

Abstract

The invention discloses a high-precision mass flow controller adjusting method based on artificial intelligence, and particularly relates to the technical field of mass flow controllers, and the method comprises the steps: collecting the recognition delay information and difference value control information of a high-precision mass flow controller, obtaining a recognition fluctuation coefficient and a control fluctuation coefficient, building a fluctuation residual value model, and obtaining a fluctuation residual value model; the method comprises the following steps: acquiring a fluctuation residual value index, performing diversity evaluation on an output signal of the high-precision mass flow controller, generating a signal composite coefficient, performing comprehensive evaluation according to the fluctuation residual value index and the signal composite coefficient, verifying the signal stability of the high-precision mass flow controller, and proposing an intervention strategy according to a verification result. According to the invention, the utilization efficiency of the artificial intelligence model is effectively improved, the accuracy and stability of the artificial intelligence model for quality flow control are improved, and the problem that a fault source is difficult to judge by a worker when a data output fault occurs is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of mass flow controllers, and more specifically, to a high-precision mass flow controller adjustment method based on artificial intelligence. Background Art

[0002] During the adjustment process of a high-precision mass flow controller with nonlinear characteristics, a slight input fluctuation of the mass flow controller may cause a drastic jump in the data obtained by the artificial intelligence algorithm, which in turn makes it difficult for the prediction model based on the artificial intelligence algorithm to learn from historical data, grasp the laws and characteristics of the mass flow controller, and effectively predict the flow change trend to achieve high-precision regulation and control. The prediction model based on artificial intelligence has a black box characteristic, and it is difficult for staff to determine whether the cause of the fault is a defect in the artificial intelligence model itself or the instability of the model input by inspecting the model data, which brings difficulties to the staff's inspection and verification.

[0003] In order to solve the above defects, a technical solution is now proposed. Summary of the invention

[0004] The purpose of the present invention is to provide a high-precision mass flow controller adjustment method based on artificial intelligence to solve the deficiencies in the background technology.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a high-precision mass flow controller adjustment method based on artificial intelligence, the specific steps include:

[0006] Collecting identification delay information and difference control information of the high-precision mass flow controller, and obtaining an identification fluctuation coefficient and a control fluctuation coefficient according to the identification delay information and the difference control information;

[0007] A fluctuation residual value model is established based on the identification fluctuation coefficient and the control fluctuation coefficient to evaluate the influence of signal identification real-time and control output fluctuation on the regulation efficiency of high-precision mass flow controller, and generate a fluctuation residual value index;

[0008] Evaluate the diversity of the output signal of the high-precision mass flow controller, obtain diversity information, and generate a signal composite coefficient based on the diversity information;

[0009] A comprehensive evaluation was performed based on the fluctuation residual index and signal composite coefficient to verify the signal stability of the high-precision mass flow controller, and an intervention strategy was proposed based on the verification results.

[0010] Preferably, the method for collecting the identification delay information of the high-precision mass flow controller and obtaining the identification fluctuation coefficient according to the identification delay information is:

[0011] Obtaining the signal recognition time of the high-precision mass flow controller within a time period T, and calibrating the signal recognition time as Tn, where n={1,2,3…m}, and m is a positive integer;

[0012] The acquisition logic of the signal recognition time of the high-precision mass flow controller within the time period T is:

[0013] Obtain the time required for the sensor to convert temperature changes into electrical signals, and calibrate the time required for the sensor to convert temperature changes into electrical signals as T1;

[0014] The time required for the acquisition electrical signal to be converted into a recognizable mass flow value is T2.

[0015] The calculation method of the signal recognition time Tn of the high-precision mass flow controller within the time period T is Tn=T1+T2, and the signal recognition time Tn of the high-precision mass flow controller within the calibration time period T is the recognition delay information;

[0016] Calculate the standard deviation Sp of the recognition delay information within the time period T, then the calculation expression of the standard deviation Sp is: Where m is the total number of identification delay information within the time period T. is the average value of the signal recognition time of the high-precision mass flow controller within the time period T, and its calculation expression is:

[0017] The recognition fluctuation coefficient is obtained according to the recognition delay information. The calculation expression of the recognition fluctuation coefficient is: vc =Sp×exp(Sp 2 +1), where I vc To identify the coefficient of volatility.

[0018] Preferably, the method for collecting the difference control information of the high-precision mass flow controller and obtaining the control fluctuation coefficient according to the difference control information is:

[0019] Obtain the effective control frequency range of the high-precision mass flow controller and calibrate the effective control frequency range of the high-precision mass flow controller as [Fr min ,Fr max ], where Fr min is the lower limit of the effective control frequency range of the high-precision mass flow controller, Fr max It is the lower limit of the effective control frequency range of the high-precision mass flow controller;

[0020] Acquire the frequency of the high-precision mass flow controller outputting the control instruction according to the difference signal, and calibrate the frequency data of the high-precision mass flow controller outputting the control instruction according to the difference signal as Pj;

[0021] The frequency of the control instruction output by the high-precision mass flow controller according to the difference signal is integrated into a data set, and the number of the frequency data is marked with j, that is, j = {1, 2, 3 ... k}, where k is a positive integer, and the data set composed of the frequency of the control instruction output by the high-precision mass flow controller according to the difference signal is calibrated as the difference control information;

[0022] Calculate the standard deviation of the difference control information, then the standard deviation In the formula, is the average value of the difference control information, and its calculation expression is

[0023] The expression for calculating the control fluctuation coefficient is: In the formula, Cv co To control the volatility.

[0024] Preferably, the method of establishing a volatility residual value model based on the identification volatility coefficient and the control volatility coefficient and generating a volatility residual value index is as follows:

[0025] After normalizing the identification volatility coefficient and the control volatility coefficient, the volatility residual value model is constructed. The expression of the volatility residual value model is: In the formula, Ab fl is the volatility residual index, α and β are the weight coefficients of identifying volatility coefficient and controlling volatility coefficient respectively, and both α and β are positive numbers.

[0026] Preferably, the output signal of the high-precision mass flow controller is evaluated for diversity, and the method for obtaining diversity information is:

[0027] Obtain the analog signal output by the high-precision mass flow controller within a time period T, and convert the analog signal z output by the high-precision mass flow controller within several consecutive time periods T into v Composed of the original data set, marked as {Z v}, where z v ∈{Z v}, where v is the number of the analog signal output by the high-precision mass flow controller within the time period T, and v = {1, 2, 3…r}, r is a positive integer, the original data set consists of the first sub-data set and the second sub-data set, when v is an odd number, the mark {Z v＝奇数} is the first sub-dataset. When v is an even number, {Z v＝偶数} is the second sub-dataset, and the diversity information includes the first sub-dataset and the second sub-dataset.

[0028] Preferably, the method for generating a signal composite coefficient according to the diversity information is:

[0029] According to the diversity information, the entropy of the first sub-dataset is calculated as H(Z v＝奇数 )=-∑[p(z v＝奇数 )*log2(p(z v＝奇数 ))], calculate the entropy of the second sub-dataset as H(Z v＝偶数 )=-∑[p(z v＝偶数 )*log2(p(z v＝偶数 ))], where H(Z v＝奇数 ) and H(Z v＝偶数 ) are the entropies of the first sub-dataset and the second sub-dataset, p(z v＝奇数 ) and p(z v＝偶数 ) are the simulated signals z in the first and second sub-datasets respectively. v The probability of taking the value of ;

[0030] The joint entropy is calculated as , where H(Z v＝奇数 ,Z v＝偶数 ) is the joint entropy, p(Z v＝奇数 ,z v＝偶数 ) represents z v＝奇数 And z v＝偶数 The joint probability of

[0031] Calculate the mutual information between the first sub-dataset and the second sub-dataset. The calculation expression is I(Z v＝奇数 ; Z v＝偶数 )=H(Z v＝奇数 )+H(Z v＝偶数 )-H(Z v＝奇数 ,Z v＝偶数 ), where I(Z v＝奇数 ; Z v＝偶数 ) is the mutual information between the first sub-dataset and the second sub-dataset, and the preset mutual information threshold is I th , the calculated mutual information I(Z v＝奇数 ; Z v＝偶数 ) and the preset mutual information threshold I th The signal composite coefficient is obtained by subtraction. The calculation expression of the signal composite coefficient is S rf =|I(Z v＝奇数 ; Z v＝偶数 )-I th |.

[0032] Preferably, the logic for verifying the signal stability of the high-precision mass flow controller is as follows:

[0033] The logic of comprehensive evaluation based on the volatility residual index and signal composite coefficient is:

[0034] Constructing a comprehensive evaluation model for Co as =γ*S rf +δ*Vr vi , where Co as is the comprehensive evaluation value, and the comprehensive evaluation threshold is set to Co th , the comprehensive evaluation threshold Co th and the comprehensive evaluation value Co as Compare, if the comprehensive evaluation value Co as Greater than or equal to the comprehensive evaluation threshold Co th , then the stability of the marking model fails;

[0035] If the comprehensive evaluation value Co as Less than the comprehensive evaluation threshold Co th , then the marking model stability takes effect.

[0036] Preferably, the logic for proposing an intervention strategy based on the verification results is:

[0037] If the model stability fails, the staff will be informed that the robustness of the artificial intelligence model is unbalanced; if the model stability is effective, the staff will be informed that the robustness of the artificial intelligence model is sufficient.

[0038] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0039] The present application obtains the identification fluctuation coefficient and the control fluctuation coefficient respectively by collecting the identification delay information and the difference control information of the high-precision mass flow controller, and performs modeling analysis based on the identification fluctuation coefficient and the control fluctuation coefficient, verifies the stability of the output data for the artificial intelligence model, and performs diversity evaluation in combination with the output signal of the high-precision mass flow controller to test the generalization effect of the output signal and the output mode for the artificial intelligence model, thereby effectively improving the utilization efficiency of the artificial intelligence model, improving the accuracy and stability of the artificial intelligence model for mass flow control, and avoiding the problem that when data output failure occurs, it is difficult for staff to determine the source of the fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0041] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] Example 1: Please refer to Figure 1 As shown, the present invention is a high-precision mass flow controller adjustment method based on artificial intelligence, and the specific steps include:

[0044] Collecting identification delay information and difference control information of the high-precision mass flow controller, and obtaining an identification fluctuation coefficient and a control fluctuation coefficient according to the identification delay information and the difference control information;

[0045] A fluctuation residual value model is established based on the identification fluctuation coefficient and the control fluctuation coefficient to evaluate the influence of signal identification real-time and control output fluctuation on the regulation efficiency of high-precision mass flow controller, and generate a fluctuation residual value index;

[0046] Perform diversity evaluation on the output signal of the high-precision mass flow controller and generate signal complex coefficients;

[0047] A comprehensive evaluation was performed based on the fluctuation residual index and signal composite coefficient to verify the signal stability of the high-precision mass flow controller, and an intervention strategy was proposed.

[0048] The high-precision mass flow controller is composed of a sensor, a regulating valve, and an amplifying control circuit component. The sensor part includes a heating element and a temperature sensor. The heating element is used to generate heat and exchange heat with the gas flowing through it. The temperature sensor is used to monitor the temperature change of the heating element. When the gas flows through the sensor part, the heating element in the sensor exchanges heat with the gas, causing the temperature of the heating element to change. The temperature sensor monitors the temperature change and converts the temperature change into an electrical signal. The electrical signal represents the mass flow of the gas. The electrical signal output by the temperature sensor is converted into a recognizable mass flow value after amplification and filtering.

[0049] The amplification, filtering, and conversion of electrical signals are all analyzed and processed by the electronic control system within the high-precision mass flow controller. The generation of electrical signals is different. Sensors based on different principles have different acquisition efficiencies for different temperature changes, and the movement patterns of the gas flowing through are also different. That is, the electronic control system has differences in the processing and conversion efficiency of different electrical signals.

[0050] If there is a delay fluctuation in converting the electrical signal into a recognizable mass flow value each time the electrical signal is acquired, then excessive delay fluctuations will have the following adverse effects on subsequent flow control and artificial intelligence model construction and prediction:

[0051] Delay fluctuations will cause the flow controller to respond more slowly and fail to adjust in time according to the current flow conditions. In situations where rapid response is required, such as real-time flow control in industrial production, delays may lead to unstable production and even safety accidents.

[0052] Delay fluctuations introduce additional errors, which reduce the adjustment accuracy of the flow controller, which may lead to unstable product quality, increase the scrap rate, and reduce production efficiency;

[0053] Latency fluctuations can cause the quality of data used for model training to deteriorate. Latency fluctuations introduce temporal inconsistencies, which weaken the correlation between data and thus affect the training effect of the model.

[0054] Due to the loss of data quality, the trained AI model cannot be effectively generalized to new data, that is, the AI ​​model’s learning effect on real-time data in actual applications is not as good as expected, and it cannot accurately predict and control traffic;

[0055] Delay fluctuations destroy the continuity of time series data, making it difficult for the AI ​​model to accurately capture the changing trends between data, resulting in reduced prediction accuracy of the AI ​​model and failure to provide a reliable prediction basis for traffic control;

[0056] Excessive delay fluctuations will cause the prediction results to lag behind the actual traffic conditions. The lag makes the prediction results of the artificial intelligence model lose practical significance and cannot provide effective support for real-time traffic control.

[0057] In order to avoid the above adverse effects, the identification delay information of the high-precision mass flow controller is collected, and the method for obtaining the identification fluctuation coefficient based on the identification delay information is as follows:

[0058] Obtaining the signal recognition time of the high-precision mass flow controller within a time period T, and calibrating the signal recognition time as Tn, where n={1,2,3…m}, and m is a positive integer;

[0059] The acquisition logic of the signal recognition time of the high-precision mass flow controller within the time period T is:

[0060] Obtain the time required for the sensor to convert temperature changes into electrical signals, and calibrate the time required for the sensor to convert temperature changes into electrical signals as T1;

[0061] The time required for the acquisition electrical signal to be converted into a recognizable mass flow value is T2.

[0062] The calculation method of the signal recognition time Tn of the high-precision mass flow controller within the time period T is Tn=T1+T2, and the signal recognition time Tn of the high-precision mass flow controller within the calibration time period T is the recognition delay information;

[0063] It should be pointed out that the time required for the sensor to convert temperature changes into electrical signals depends on the type and performance of the temperature sensor, which can be obtained by professional and technical personnel in this field through actual testing. The time required for the electrical signal to be converted into a recognizable mass flow value is determined by the processing speed and algorithm complexity of the electronic control system, which can be obtained by performing performance tests on the electronic control system or consulting the technical specifications.

[0064] Calculate the standard deviation Sp of the recognition delay information within the time period T, then the calculation expression of the standard deviation Sp is: Where m is the total number of identification delay information within the time period T. is the average value of the signal recognition time of the high-precision mass flow controller within the time period T, and its calculation expression is:

[0065] The recognition fluctuation coefficient is obtained according to the recognition delay information. The calculation expression of the recognition fluctuation coefficient is: vc =Sp×exp(Sp 2 +1), where I vc To identify the coefficient of volatility.

[0066] After obtaining the recognizable mass flow value, the recognizable mass flow value is compared with a preset mass flow threshold through the electronic control system to obtain a difference signal between the recognizable mass flow value and the preset mass flow threshold. The electronic control system sends a control signal to the regulating valve according to the difference signal to accurately control the gas flow;

[0067] If the frequency of sending control instructions based on the difference signal fluctuates too much, the excessive fluctuation will have the following adverse effects on flow control and artificial intelligence model construction and prediction:

[0068] The fluctuation of command frequency will lead to unstable opening adjustment of flow control valve, thus causing fluctuation of actual flow. Unstable flow control directly affects the stability of production process and consistency of product quality.

[0069] When the command frequency fluctuates too much, it may not be possible to respond and adjust the flow in time, resulting in control lag, which may cause production accidents or equipment damage in flow control that requires fast response;

[0070] Frequent command fluctuations and valve opening adjustments will increase equipment wear and fatigue, which will not only reduce the life of the equipment, but may also increase maintenance costs and downtime;

[0071] Fluctuations in instruction frequency will lead to a decrease in the quality of data used for AI model training. The instability and noise in the data will affect the training effect of the AI ​​model, reducing the accuracy and generalization ability of the AI ​​model.

[0072] When data quality decreases, the training process of AI models may become more difficult and complicated, and the data volume and training time requirements for AI models will be further increased;

[0073] The fluctuation of instruction frequency will destroy the continuity and stability of data, making it difficult for the artificial intelligence model to accurately capture the changing trend between data, resulting in reduced prediction accuracy of the model and failure to provide reliable prediction basis for flow control;

[0074] When the instruction frequency fluctuates too much, the prediction results of the artificial intelligence model may be delayed. The delay may make the prediction results lose practical significance and fail to provide effective support for real-time traffic control.

[0075] In order to avoid the above adverse effects, the difference control information of the high-precision mass flow controller is collected, and the method for obtaining the control fluctuation coefficient according to the difference control information is as follows:

[0076] Obtain the effective control frequency range of the high-precision mass flow controller and calibrate the effective control frequency range of the high-precision mass flow controller as [Fr min ,Fr max ], where Fr min is the lower limit of the effective control frequency range of the high-precision mass flow controller, Fr max It is the lower limit of the effective control frequency range of the high-precision mass flow controller;

[0077] It should be noted that the upper and lower limits of the effective control frequency of the high-precision mass flow controller can be determined by performing performance tests on the high-precision mass flow controller and observing its control stability and accuracy at different frequencies, or by referring to the technical specifications and related literature of the high-precision mass flow controller.

[0078] Acquire the frequency of the high-precision mass flow controller outputting the control instruction according to the difference signal, and calibrate the frequency data of the high-precision mass flow controller outputting the control instruction according to the difference signal as Pj;

[0079] The frequency of the control instruction output by the high-precision mass flow controller according to the difference signal is integrated into a data set, and the number of the frequency data is marked with j, that is, j = {1, 2, 3 ... k}, where k is a positive integer, and the data set composed of the frequency of the control instruction output by the high-precision mass flow controller according to the difference signal is calibrated as the difference control information;

[0080] Calculate the standard deviation of the difference control information, then the standard deviation In the formula, is the average value of the difference control information, and its calculation expression is

[0081] The expression for calculating the control fluctuation coefficient is: In the formula, Cv co To control the volatility.

[0082] A fluctuation residual value model is established based on the identification fluctuation coefficient and the control fluctuation coefficient to evaluate the influence of signal identification real-time and control output fluctuation on the regulation efficiency of high-precision mass flow controller, and generate a fluctuation residual value index;

[0083] The method of establishing a volatility residual value model based on the identification volatility coefficient and the control volatility coefficient and generating a volatility residual value index is as follows:

[0084] After normalizing the identification volatility coefficient and the control volatility coefficient, the volatility residual value model is constructed. The expression of the volatility residual value model is: In the formula, Ab fl is the volatility residual index, α and β are the weight coefficients of identifying volatility coefficient and controlling volatility coefficient respectively, and both α and β are positive numbers.

[0085] Evaluate the diversity of the output signal of the high-precision mass flow controller, obtain diversity information, and generate a signal composite coefficient based on the diversity information;

[0086] Obtain the analog signal output by the high-precision mass flow controller within a time period T, and convert the analog signal z output by the high-precision mass flow controller within several consecutive time periods T into v Composed of the original data set, marked as {Z v}, where z v ∈{Z v}, where v is the number of the analog signal output by the high-precision mass flow controller within the time period T, and v = {1, 2, 3…r}, r is a positive integer, the original data set consists of the first sub-data set and the second sub-data set, when v is an odd number, the mark {Z v＝奇数} is the first sub-dataset. When v is an even number, {Z v＝偶数} is the second sub-dataset, and the diversity information includes the first sub-dataset and the second sub-dataset;

[0087] The method for generating signal composite coefficients based on diversity information is:

[0088] According to the diversity information, the entropy of the first sub-dataset is calculated as H(Z v＝奇数 )=-∑[p(z v＝奇数 )*log2(p(z v＝奇数 ))], calculate the entropy of the second sub-dataset as H(Z v＝偶数 )=-∑[p(z v＝偶数 )*log2(p(z v＝偶数 ))], where H(Z v＝奇数 ) and H(Z v＝偶数 ) are the entropies of the first sub-dataset and the second sub-dataset, p(z v＝奇数 ) and p(z v＝偶数 ) are the simulated signals z in the first and second sub-datasets respectively. v The probability of taking the value of ;

[0089] The joint entropy is calculated as , where H(Z v＝奇数 ,Z v＝偶数 ) is the joint entropy, p(z v＝奇数 ,z v＝偶数 ) represents z v＝奇数 And z v＝偶数 The joint probability of

[0090] Calculate the mutual information between the first sub-dataset and the second sub-dataset. The calculation expression is I(Z v＝奇数 ; Z v＝偶数 )=H(Z v＝奇数 )+H(Z v＝偶数 )-H(Z v＝奇数 ,Z v＝偶数 ), where I(Z v＝奇数 ; Z v＝偶数 ) is the mutual information between the first sub-dataset and the second sub-dataset, and the preset mutual information threshold is I th , the calculated mutual information I(Z v＝奇数 ; Z v＝偶数 ) and the preset mutual information threshold I th The signal composite coefficient is obtained by subtraction. The calculation expression of the signal composite coefficient is S rf =|I(Z v＝奇数 ; Z v＝偶数 )-I th |.

[0091] It should be noted that the mutual information threshold I thIt is set by professional and technical personnel in this field according to the output performance of the high-precision mass flow controller, which will not be elaborated here. The analog signals output by the high-precision mass flow controller within several time periods T are cross-classified and evaluated according to the odd and even number classification, which effectively improves the accuracy of mutual information calculation and improves the evaluation efficiency of the diversity of output signals within the cycle time T.

[0092] Comprehensive evaluation was performed based on the fluctuation residual index and signal composite coefficient to verify the signal stability of the high-precision mass flow controller and propose intervention strategies;

[0093] The logic of comprehensive evaluation based on the volatility residual index and signal composite coefficient is:

[0094] Constructing a comprehensive evaluation model for Co as =γ*S rf +δ*Vr vi , where Co as is the comprehensive evaluation value, and the comprehensive evaluation threshold is set to Co th , the comprehensive evaluation threshold Co th and the comprehensive evaluation value Co as For comparison, if the comprehensive evaluation value Co as Greater than or equal to the comprehensive evaluation threshold Co th , then the model stability is marked as failed, and the staff is informed that the robustness of the artificial intelligence model is unbalanced;

[0095] If the comprehensive evaluation value Co as Less than the comprehensive evaluation threshold Co th , then the model stability mark is effective, informing the staff that the AI ​​model is sufficiently robust.

[0096] The present application obtains the identification fluctuation coefficient and the control fluctuation coefficient respectively by collecting the identification delay information and the difference control information of the high-precision mass flow controller, and performs modeling analysis based on the identification fluctuation coefficient and the control fluctuation coefficient, verifies the stability of the output data for the artificial intelligence model, and performs diversity evaluation in combination with the output signal of the high-precision mass flow controller to test the generalization effect of the output signal and the output mode for the artificial intelligence model, thereby effectively improving the utilization efficiency of the artificial intelligence model, improving the accuracy and stability of the artificial intelligence model for mass flow control, and avoiding the problem that when data output failure occurs, it is difficult for staff to determine the source of the fault.

[0097] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0098] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of computer program goods. The computer program goods include one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0099] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0100] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0102] If the functions are implemented in the form of software functional units and sold or used as independent goods, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of software goods, which are stored in a storage medium and include several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0103] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A high-precision mass flow controller adjustment method based on artificial intelligence, characterized in that: The specific steps include: Collecting identification delay information and difference control information of the high-precision mass flow controller, and obtaining an identification fluctuation coefficient and a control fluctuation coefficient according to the identification delay information and the difference control information; A fluctuation residual value model is established based on the identification fluctuation coefficient and the control fluctuation coefficient to evaluate the influence of signal identification real-time and control output fluctuation on the regulation efficiency of high-precision mass flow controller, and generate a fluctuation residual value index; Evaluate the diversity of the output signal of the high-precision mass flow controller, obtain diversity information, and generate a signal composite coefficient based on the diversity information; A comprehensive evaluation was performed based on the fluctuation residual index and signal composite coefficient to verify the signal stability of the high-precision mass flow controller, and an intervention strategy was proposed based on the verification results.

2. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 1 is characterized in that: The method for collecting the identification delay information of the high-precision mass flow controller and obtaining the identification fluctuation coefficient according to the identification delay information is as follows: Obtaining the signal recognition time of the high-precision mass flow controller within a time period T, and calibrating the signal recognition time as Tn, where n={1,2,3…m}, and m is a positive integer; The acquisition logic of the signal recognition time of the high-precision mass flow controller within the time period T is: Obtain the time required for the sensor to convert temperature changes into electrical signals, and calibrate the time required for the sensor to convert temperature changes into electrical signals as T1; The time required for the acquisition electrical signal to be converted into a recognizable mass flow value is T2. The calculation method of the signal recognition time Tn of the high-precision mass flow controller within the time period T is Tn=T1+T2, and the signal recognition time Tn of the high-precision mass flow controller within the calibration time period T is the recognition delay information; Calculate the standard deviation Sp of the recognition delay information within the time period T, then the calculation expression of the standard deviation Sp is: Where m is the total number of identification delay information within the time period T. is the average value of the signal recognition time of the high-precision mass flow controller within the time period T, and its calculation expression is: The recognition fluctuation coefficient is obtained according to the recognition delay information. The calculation expression of the recognition fluctuation coefficient is: vc =Sp×exp(Sp 2 +1), where I vc To identify the coefficient of volatility.

3. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 1 is characterized in that: The method for collecting the differential control information of the high-precision mass flow controller and obtaining the control fluctuation coefficient according to the differential control information is as follows: Obtain the effective control frequency range of the high-precision mass flow controller and calibrate the effective control frequency range of the high-precision mass flow controller as [Fr min ,[Fr max ], where Fr min is the lower limit of the effective control frequency range of the high-precision mass flow controller, Fr max It is the lower limit of the effective control frequency range of the high-precision mass flow controller; Acquire the frequency of the high-precision mass flow controller outputting the control instruction according to the difference signal, and calibrate the frequency data of the high-precision mass flow controller outputting the control instruction according to the difference signal as Pj; The frequency of the control instruction output by the high-precision mass flow controller according to the difference signal is integrated into a data set, and the number of the frequency data is marked with j, that is, j = {1, 2, 3 ... k}, where k is a positive integer, and the data set composed of the frequency of the control instruction output by the high-precision mass flow controller according to the difference signal is calibrated as the difference control information; Calculate the standard deviation of the difference control information, then the standard deviation In the formula, is the average value of the difference control information, and its calculation expression is The expression for calculating the control fluctuation coefficient is: Where, Cv co To control the volatility.

4. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 3 is characterized in that: The method of establishing a volatility residual value model based on the identification volatility coefficient and the control volatility coefficient and generating a volatility residual value index is as follows: After normalizing the identification volatility coefficient and the control volatility coefficient, the volatility residual value model is constructed. The expression of the volatility residual value model is: In the formula, Ab fl is the volatility residual index, α and β are the weight coefficients of identifying volatility coefficient and controlling volatility coefficient respectively, and both α and β are positive numbers.

5. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 1 is characterized in that: The output signal of the high-precision mass flow controller is evaluated for diversity, and the method for obtaining diversity information is as follows: Obtain the analog signal output by the high-precision mass flow controller within a time period T, and convert the analog signal z output by the high-precision mass flow controller within several consecutive time periods T into v Composed of the original data set, marked as {Z v }, where z v ∈{Z v }, where v is the number of the analog signal output by the high-precision mass flow controller within the time period T, and v = {1, 2, 3…r}, r is a positive integer, the original data set consists of the first sub-data set and the second sub-data set, when v is an odd number, the mark {Z v＝奇数 } is the first sub-dataset. When v is an even number, {Z v＝偶数 } is the second sub-dataset, and the diversity information includes the first sub-dataset and the second sub-dataset.

6. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 5 is characterized in that: The method for generating signal composite coefficients based on diversity information is: According to the diversity information, the entropy of the first sub-dataset is calculated as H(Z v＝奇数 )=-∑[p(z v＝奇数 )*log2(p(z v＝奇数 ))], calculate the entropy of the second sub-dataset as H(z v＝偶数 )=-∑[p(z v＝偶数 )*log2(p(z v＝偶数 ))], where H(Z v＝奇数 ) and H(Z v＝偶数 ) are the entropies of the first sub-dataset and the second sub-dataset, p(z v＝奇数 ) and p(z v＝偶数 ) are the simulated signals z in the first and second sub-datasets respectively. v The probability of taking the value of ; The joint entropy is calculated as , where H(Z v＝奇数 ,Z v＝偶数 ) is the joint entropy, p(z v＝奇数 ,z v＝偶数 ) represents z v＝奇数 And z v＝偶数 The joint probability of Calculate the mutual information between the first sub-dataset and the second sub-dataset. The calculation expression is I(Z v＝奇数 ; Z v＝偶数 )=H(Z v＝奇数 )+H(Z v＝偶数 )-H(Z v＝奇数 ,Z v＝偶数 ), where I(Z v＝奇数 ; Z v＝偶数 ) is the mutual information between the first sub-dataset and the second sub-dataset, and the preset mutual information threshold is I th , the calculated mutual information I(Z v＝奇数 ; Z v＝偶数 ) and the preset mutual information threshold I th The signal composite coefficient is obtained by subtraction. The calculation expression of the signal composite coefficient is S rf =|I(Z v＝奇数 ; Z v＝偶数 )-I th |.

7. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 6 is characterized in that: The logic for verifying the signal stability of the high-precision mass flow controller is as follows: The logic of comprehensive evaluation based on the volatility residual index and signal composite coefficient is: Constructing a comprehensive evaluation model for Co as =γ*S rf +δ*Vr vi , where Co as is the comprehensive evaluation value, and the comprehensive evaluation threshold is set to Co th , the comprehensive evaluation threshold Co th and the comprehensive evaluation value Co as For comparison, if the comprehensive evaluation value Co as Greater than or equal to the comprehensive evaluation threshold Co th , then the stability of the marking model fails; If the comprehensive evaluation value Co as Less than the comprehensive evaluation threshold Co th , then the marking model stability takes effect.

8. The high-precision mass flow controller adjustment method based on artificial intelligence according to claim 7 is characterized in that: The logic of the intervention strategy proposed based on the verification results is: If the model stability fails, the staff will be informed that the robustness of the artificial intelligence model is unbalanced; if the model stability is effective, the staff will be informed that the robustness of the artificial intelligence model is sufficient.

Images