Nonlinear evaluation method of closed loop based on pid parameter identification

Abstract

The embodiment of the present application provides a kind of closed loop circuit nonlinear evaluation method based on PID parameter identification, belong to the field of petrochemical industry.The evaluation method includes: according to the operation data of the plurality of sampling time, the deviation sequence corresponding to sampling time is obtained;The deviation sequence is applied at least one discrete control model, obtains the model reference parameter set corresponding to each discrete control model, each reference parameter in the model reference parameter set forms reference parameter sequence according to the order of sampling time;Using parameter identification probability evaluation model, all reference parameter sequences are identified probability analysis, and the identification rate of all reference parameter sequences is obtained;If the identification rate of all reference parameter sequences corresponding to model reference parameter set is greater than preset value, and the maximum identification rate in the all identification rate corresponding to the discrete control model is determined the nonlinear degree of the control loop.The present application is conducive to linear control model modeling and fault diagnosis of regulating valve.

Description

Technical Field

[0001] This invention relates to the technical field of control system safety in petrochemical industry, and more specifically, to a closed-loop nonlinear evaluation method based on PID parameter identification. Background Technology

[0002] The performance degradation of control loops is a common problem in the control systems of refining and chemical plants. Nonlinear characteristics within these loops are a major contributing factor, severely impacting the safety and economic efficiency of the plants. Nonlinear detection techniques typically rely on statistical hypothesis testing, and based on the selection of statistical indicators, they can be categorized into parametric methods and cost-effective methods. For example, Choudhury proposed a nonlinear detection method based on biphase coherence spectrum, but this method has limitations regarding data length and is not highly accurate for weak nonlinearities. A control loop nonlinear detection method based on improved inverted biphase spectrum analysis, compared to Choudhury's biphase coherence spectrum method, has less dependence on the length of the measured data and higher accuracy for weak nonlinearities. An industrial process nonlinear detection method based on substitution data constructs multiple sets of substitution data for the original signal using the LMD-RP substitution data method, and calculates the inverted biphase coherence spectrum function of each set of substitution data, mimicking the original signal. A nonlinear identification method for main steam regulating valves based on the Bottom-Up algorithm uses the least squares method to identify the parameters of each linear model, and calculates the loss function for each fusion interval based on the parameter identification results.

[0003] However, the above methods are all based on statistical methods of higher-order spectra for nonlinear identification, which have requirements on the amount and quality of data. When the amount of data is too small, the above methods cannot correctly identify nonlinearity. Summary of the Invention

[0004] The purpose of this invention is to provide a closed-loop nonlinearity evaluation method based on PID parameter identification, which mainly solves the problem of identifying the nonlinearity of operating parameters in the closed-loop control process.

[0005] To achieve the above objectives, embodiments of the present invention provide a closed-loop nonlinearity evaluation method based on PID parameter identification, the nonlinearity evaluation method comprising:

[0006] The operation data of the control loop at multiple sampling times are obtained, and the deviation data corresponding to the sampling times are obtained based on the operation data at the multiple sampling times and a deviation sequence is formed according to the order of the sampling times.

[0007] The deviation sequence is applied to at least one discrete control model to obtain a model reference parameter set corresponding to each discrete control model. Each reference parameter in the model reference parameter set forms a reference parameter sequence according to the order of sampling time.

[0008] The identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate of all reference parameter sequences;

[0009] If the identification rate of all reference parameter sequences corresponding to the model reference parameter set is greater than the preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop, and the nonlinearity of the control loop is determined according to the maximum identification rate among all identification rates corresponding to the discrete control model of the control loop.

[0010] Optionally, the nonlinear evaluation method further includes:

[0011] The identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate and identification value corresponding to each reference parameter sequence; the identification rate and identification value of each reference parameter sequence correspond to each other;

[0012] If the identification rate of all reference parameter sequences corresponding to the model reference parameter set is greater than the preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop. The maximum identification rate and the identification value corresponding to the maximum identification rate are determined among all the identification rates of the discrete control model of the control loop, and the reference parameter sequence corresponding to the maximum identification rate is determined.

[0013] Based on the reference parameter sequence corresponding to the maximum recognition rate and the maximum recognition rate, the linear operating probability time series of the control loop is determined.

[0014] Optionally, the nonlinear evaluation method further includes:

[0015] The linear operating probability time series of the control loop is continuously grouped according to a preset number of intervals to obtain a continuously grouped sequence;

[0016] The marker for each sequence group is determined by the magnitude of the identification value corresponding to each group in the continuous grouping sequence;

[0017] The linear operating probability parameter set sequence of the control loop is determined based on the types of any two consecutive markers.

[0018] Optionally, the markers include non-linear markers and linear markers; the markers for determining each set of sequences include:

[0019] If any identification value in the corresponding sequence group is less than the second preset value, the sequence group is marked with a non-linear marker; otherwise, the sequence group is marked with a linear marker.

[0020] The step of determining whether the sequence of linear operating probability parameters of the control loop is a linear sequence based on the types of any two consecutive markers includes:

[0021] If any two consecutive sets of sequences have markers of the type of non-linear markers and linear markers respectively, or if any two consecutive sets of sequences have markers of the type of linear markers and non-linear markers respectively, then the two consecutive sets of sequences are determined to be linear sequences.

[0022] Optionally, the step of using a parameter identification probability evaluation model to perform identification probability analysis on all reference parameter sequences to obtain the identification rate and identification value corresponding to all reference parameter sequences, wherein the identification rate and identification value of each reference parameter sequence correspond to each other, including:

[0023] S1) Determine the number of equal segments SPCount in the value range of the identification sequence B; wherein the identification sequence B is a reference parameter sequence;

[0024] S2) Determine the number of data items nCount in the subset newB of the identification sequence B; the data item size of the subset newB is equal to the data item size of the identification sequence B;

[0025] S3) Determine the data segmentation interval sp based on the maximum and minimum values ​​in the subset newB and the number of segments SPCount;

[0026] S4) Based on the number of segments SPCount and the data segmentation interval sp, initialize the segmentation interval sequence sps and the amount of data within the data segmentation interval sp;

[0027] S5) Based on the value of subset newB, determine the number of data items spcount whose values ​​conform to the range of each data partition interval sp;

[0028] S6) Obtain the maximum value spmax from the number of data items spcount;

[0029] S7) Remove data from other segmentation periods whose data volume is less than a preset value, store the removed subset newB and update the data volume of the subset newB; the preset value is a preset multiple of the maximum value spmax.

[0030] S8) Repeat steps S2)-S7) until the number of iterations reaches the set value or the data size of subset newB is less than the number of splits SPCount. Then, output the data size of subset newB as the data size of the result subset newB and stop the loop.

[0031] S9) Obtain the average value avg of the data volume of the result subset newB, and select the subset newB whose data volume value and average value avg satisfy the preset condition as the selected subset newB;

[0032] S10) Obtain the average value avg of all selected subsets newB. Calculate the similarity S between the value of selected subset newB and the average value avg of selected subset newB, and obtain the average value Savg of the similarity S between the value of selected subset newB and the average value avg of selected subset newB.

[0033] S11) Based on the average value Savg of the similarity degree S, the value of the selected subset newB, and the amount of data in the selected subset newB, calculate the recognition rate and recognition value corresponding to the amount of data in the selected subset newB, and use them as the recognition rate and recognition value corresponding to the reference parameter sequence, respectively.

[0034] Optionally, the recognition rate and recognition value corresponding to the data volume of the selected subset newB can be calculated using the following formula:

[0035] B rely =Savg×nCount / AllCount;

[0036] B avg =∑newB n / nCount;

[0037] Among them, B rely For recognition rate; B avg For identification value; newB n The data size of the selected subset newB is ; AIICount is the number of data items in the identification sequence B.

[0038] Optionally, the model reference parameter set includes one or more of the following: gain parameter K, integral time parameter T, and derivative time parameter D.

[0039] Optionally, the step of acquiring the operating data of the control loop at multiple sampling times, and obtaining deviation data corresponding to the sampling times based on the operating data at the multiple sampling times and forming a deviation sequence, includes:

[0040] Acquire the setpoint SP, measured value PV, and output value MV of the control loop at multiple sampling times;

[0041] The following deviation calculation model is used to calculate the deviation data corresponding to the sampling time and form a deviation sequence;

[0042] E n =PV n -SP nn = 1…4800

[0043] dE n =E n -E n-1 n = 2…4800

[0044] dPV n =PV n -PV n-1 n = 2…4800

[0045] dMV n =MV n -MV n-1 n = 2…4800

[0046] ddE n =dE n -dE n-1 n = 3…4800

[0047] ddPV n =dPV n -dPV n-1 n = 3…4800

[0048] Among them, PV n SP is the PV value at time n; n E is the SP value at time n; n dE is the deviation E between the measured value PV and the set value SP at time n. n Let dE be the deviation value of the deviation E at time n. When n is less than 2, dE is... n =0; dPV n dPV is the deviation of the PV value at time n, where n is less than 2. n =0; dMV n dMV is the deviation of the MV value at time n, where n is less than 2. n =0;ddE n ddE is the deviation of dE at time n. When n is less than 3, ddE n =0; ddPV n The deviation of dPV at time n is the deviation value when n is less than 3. n =0.

[0049] Optionally, the discrete control model includes one or more of the following: a deviation-based proportional-integral-derivative (PID) control model, a deviation-based proportional-integral (PI) control model, a deviation-based proportional control (P) model, a deviation-based integral control (I) model, a measurement-based derivative and deviation-based proportional-integral (PI-D) control model, and a measurement-based proportional-derivative and deviation-based integral control (I-PD) model.

[0050] Alternatively, the deviation-based proportional-integral-derivative (PID) control model is as follows:

[0051] ΔMV n =K×ΔE n +T*E n +D×Δ(ΔE n );

[0052] K = K p ×K s ;

[0053]

[0054]

[0055] k p =100 / PB;

[0056] K s = (MSH-MSL) / (SH-SL);

[0057] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable; E n ΔE represents the deviation between the measured value and the set value at time n. n E at time n n E at time n-1 n The deviation; Δ(ΔE) n ) represents ΔE at the nth time. n ΔE at time n-1 n Deviation; MV n The output value at time n; ΔMV n MV at time n n MV at time n-1 n The deviation.

[0058] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0059] The gain parameter K, integral time parameter T, and derivative time parameter D of the deviation-based proportional-integral-derivative (PID) control are obtained by solving the deviation-based PID model and the deviation calculation model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; Dn is the calculation formula for the derivative time parameter D at time n; where:

[0060]

[0061]

[0062]

[0063] In the above formula: a1=dEsn-2, a2=dEsn-1, a3=dEsn, b1=Esn-2, b2=Esn-1, b3=Esn, c1=ddEsn-2, c2=ddEsn-1, c3=ddEsn, d1=dMvsn-2, d2=dMvsn-1, d3=dMvsn; n is the time position of the time series.

[0064] Alternatively, the deviation-based proportional-integral (PI) control model is:

[0065] ΔMV n =K×ΔE n +T×E n ;

[0066] K = K p ×K s ;

[0067]

[0068] K p =100 / PB;

[0069] K s = (MSH-MSL) / (SH-SL);

[0070] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable.

[0071] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0072] The gain parameter K and integral time parameter T corresponding to the deviation-based proportional-integral (PI) control are obtained by solving the deviation-based PI model and the deviation calculation model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; where:

[0073]

[0074]

[0075] Alternatively, the deviation-based proportional control (P) model is:

[0076] ΔMV n =K×ΔE n ;

[0077] K = K p ×K s ;

[0078] K p =100 / PB;

[0079] K s = (MSH-MSL) / (SH-SL);

[0080] Where Kp is the process gain, PB is the proportional gain, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable.

[0081] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0082] The gain parameter K corresponding to the bias-based proportional control (P) model is obtained by solving the bias-based proportional control (P) model and the bias calculation model; Kn is the calculation formula for the gain parameter K at the nth time step; where:

[0083]

[0084] Alternatively, the deviation-based integral control (I) model is:

[0085] ΔMV n =T×E n :

[0086]

[0087] K p =100 / PB;

[0088] K s = (MSH-MSL) / (SH-SL);

[0089] Where Kp is the process gain, PB is the proportional gain, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable.

[0090] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0091] The integral time parameter T corresponding to the deviation-based integral control (I) model is obtained by solving the deviation-based integral control (I) model and the deviation calculation model; Tn is the calculation formula for the integral time parameter T at the nth time step; where:

[0092]

[0093] Optionally, the proportional-integral control (PI-D) model based on the derivative of the measured value and the deviation is as follows:

[0094] ΔMV n =K×ΔE n +T×E n +D×Δ(ΔPV n );

[0095]

[0096] K = K p ×K s ;

[0097]

[0098] K p =100 / PB;

[0099] K s = (MSH-MSL) / (SH-SL);

[0100] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, MSL is the lower limit of the manipulated variable; PV n The measurement value at time n; ΔPV n For the PV at time n n PV at time n-1 n The deviation; Δ(ΔPV) n ) represents the ΔPV at time n. n ΔPV at time n-1 n Deviation; MV n The output value at time n; ΔMV n MV at time n n MV at time n-1 n The deviation.

[0101] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0102] Solving the proportional-integral-d (PI-D) control model based on the derivative and deviation of the measured values, along with the deviation calculation model, yields the gain parameter K, integral time parameter T, and derivative time parameter D corresponding to the PI-D control based on the derivative and deviation of the measured values; Kn is the calculation formula for the gain parameter K at the nth time; Tn is the calculation formula for the integral time parameter T at the nth time; Dn is the calculation formula for the derivative time parameter D at the nth time; where:

[0103]

[0104]

[0105]

[0106] In the above formula: a1=dEsn-2, a2=dEsn-1, a3=dEsn, b1=Esn-2, b2=Esn-1, b3=Esn, c1=ddPVsn-2, c2=ddPVsn-1, c3=ddPVsn, d1=dMvsn-2, d2=dMvsn-1, d3=dMvsn; n is the time position of the time series.

[0107] Optionally, the proportional-derivative model based on the measured values ​​and the integral-PD model based on the deviation are:

[0108] ΔMV n =K×ΔPV n +T×E n +D×Δ(ΔPV n );

[0109]

[0110] K = K p ×K s ;

[0111]

[0112] K p =100 / PB;

[0113] K s = (MSH-MSL) / (SH-SL);

[0114] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, MSL is the lower limit of the manipulated variable; PV n The measurement value at time n; ΔPV nFor the PV at time n n PV at time n-1 n The deviation; Δ(ΔPV) n ) represents the ΔPV at time n. n ΔPV at time n-1 n Deviation; MV n The output value at time n; ΔMV n MV at time n n MV at time n-1 n The deviation.

[0115] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0116] Solving the measured value-based proportional-derivative model, the deviation integral control (I-PD) model, and the deviation calculation model yields the gain parameter K, integral time parameter T, and derivative time parameter D corresponding to the measured value-based proportional-derivative model and the deviation integral control (I-PD) model; Kn is the calculation formula for the gain parameter K at the nth time; Tn is the calculation formula for the integral time parameter T at the nth time; Dn is the calculation formula for the derivative time parameter D at the nth time; where:

[0117]

[0118]

[0119]

[0120] In the above formula: a1=dPVsn-2, a2=dPVsn-1, a3=dPVsn, b1=Esn-2, b2=Esn-1, b3=Esn, c1=ddPVsn-2, c2=ddPVsn-1, c3=ddPVsn, d1=dMvsn-2, d2=dMvsn-1, d3=dMvsn; n is the time position of the time series.

[0121] The present invention also provides a machine-readable storage medium storing instructions for causing a machine to execute the above-described closed-loop nonlinear evaluation method based on PID parameter identification.

[0122] Through the above technical solution, the present invention identifies PID parameters. Based on the identification probability of PID parameters, when the identification rate of the model reference parameter set corresponding to the discrete control model is greater than the preset value, the nonlinearity of the control loop is determined according to the maximum identification rate. This can be beneficial for linear control model modeling and fault diagnosis of regulating valves.

[0123] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0124] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0125] Figure 1 This is a schematic diagram of a closed-loop nonlinear evaluation method based on PID parameter identification provided in an embodiment of this application.

[0126] Figure 2 This is a schematic diagram of another closed-loop nonlinear evaluation method based on PID parameter identification provided in the embodiments of this application;

[0127] Figure 3 This is a schematic diagram of another closed-loop nonlinear evaluation method based on PID parameter identification provided in the embodiments of this application;

[0128] Figure 4 It is a historical operating trend chart of the control deviation, valve output, and linear sequence of the LIC_20702 test. Detailed Implementation

[0129] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0130] like Figure 1 As shown in this embodiment, the present invention provides a closed-loop nonlinearity evaluation method based on PID parameter identification, the nonlinearity evaluation method comprising:

[0131] The operation data of the control loop at multiple sampling times are obtained, and the deviation data corresponding to the sampling times are obtained based on the operation data at the multiple sampling times and a deviation sequence is formed according to the order of the sampling times.

[0132] The deviation sequence is applied to at least one discrete control model to obtain a model reference parameter set corresponding to each discrete control model. Each reference parameter in the model reference parameter set forms a reference parameter sequence according to the order of sampling time.

[0133] The identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate of all reference parameter sequences;

[0134] If the identification rate of all reference parameter sequences corresponding to the model reference parameter set is greater than the preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop, and the nonlinearity of the control loop is determined according to the maximum identification rate among all identification rates corresponding to the discrete control model of the control loop.

[0135] like Figure 2 As shown, in another possible embodiment, a closed-loop nonlinearity evaluation method based on PID parameter identification includes:

[0136] The operation data of the control loop at multiple sampling times are obtained, and the deviation data corresponding to the sampling times are obtained based on the operation data at the multiple sampling times and a deviation sequence is formed according to the order of the sampling times.

[0137] The deviation sequence is applied to at least one discrete control model to obtain a model reference parameter set corresponding to each discrete control model. Each reference parameter in the model reference parameter set forms a reference parameter sequence according to the order of sampling time.

[0138] The identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate and identification value corresponding to each reference parameter sequence; the identification rate and identification value of each reference parameter sequence correspond to each other;

[0139] If the identification rate of all reference parameter sequences corresponding to the model reference parameter set is greater than the preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop. The maximum identification rate and the identification value corresponding to the maximum identification rate are determined among all the identification rates of the discrete control model of the control loop, and the reference parameter sequence corresponding to the maximum identification rate is determined.

[0140] Based on the reference parameter sequence corresponding to the maximum recognition rate and the maximum recognition rate, the linear operating probability time series of the control loop is determined.

[0141] like Figure 3 As shown, the nonlinear evaluation method further includes:

[0142] The linear operating probability time series of the control loop is continuously grouped according to a preset number of intervals to obtain a continuously grouped sequence;

[0143] The marker for each sequence group is determined by the magnitude of the identification value corresponding to each group in the continuous grouping sequence;

[0144] The linear operating probability parameter set sequence of the control loop is determined based on the types of any two consecutive markers.

[0145] Optionally, the step of acquiring the operating data of the control loop at multiple sampling times, and obtaining deviation data corresponding to the sampling times based on the operating data at the multiple sampling times and forming a deviation sequence, includes:

[0146] The system acquires the setpoint SP, measured value PV, and output value MV of the control loop at multiple sampling times. Specifically, it collects the operating data of the control loop in real time through an OPC Client. This operating data includes the setpoint SP, measured value PV, output value MV, and control mode MODE. The discrete PID calculation module of the DCS control loop typically operates on a 1-second control cycle. This embodiment uses a nonlinear evaluation method based on PID parameter identification after time misalignment, so the sampling period should be greater than 1 second, preferably between 3 and 10 seconds. This embodiment defines a data sampling period of 3 seconds, and the acquisition device collects 4800 data points from each loop over 4 hours for nonlinear evaluation. Based on the MODE value, it determines whether the mode is manual or automatic. If it is manual, the setpoint SP, measured value PV, and output value MV at that time are discarded, and the setpoint SP, measured value PV, and output value MV under automatic conditions are preferred.

[0147] The following deviation calculation model is used to calculate the deviation data corresponding to the sampling time and form a deviation sequence;

[0148] E n =PV n -SP n n = 1…4800

[0149] dE n =E n -E n-1 n = 2…4800

[0150] dPV n =PV n -PV n-1 n = 2…4800

[0151] dMV n =MV n -MV n-1 n = 2…4800

[0152] ddE n =dE n -dE n-1 n = 3…4800

[0153] ddPV n =dPV n -dPV n-1 , n = 3…4800.

[0154] Among them, PV nSP is the PV value at time n; n E is the SP value at time n; n dE is the deviation E between the measured value PV and the set value SP at time n. n Let dE be the deviation value of the deviation E at time n. When n is less than 2, dE is... n =0; dPV n dPV is the deviation of the PV value at time n, where n is less than 2. n =0; dMV n dMV is the deviation of the MV value at time n, where n is less than 2. n =0;ddE n ddE is the deviation of dE at time n. When n is less than 3, ddE n =0; ddPV n The deviation of dPV at time n is the deviation value when n is less than 3. n =0.

[0155] Optionally, the discrete control model includes one or more of the following: a deviation-based proportional-integral-derivative (PID) control model, a deviation-based proportional-integral (PI) control model, a deviation-based proportional control (P) model, a deviation-based integral control (I) model, a measurement-based derivative and deviation-based proportional-integral (PI-D) control model, and a measurement-based proportional-derivative and deviation-based integral control (I-PD) model.

[0156] Optionally, the model reference parameter set includes one or more of the following: gain parameter K, integral time parameter T, and derivative time parameter D.

[0157] In one possible embodiment, the deviation-based proportional-integral-derivative (PID) control model is as follows:

[0158] ΔMV n =K×ΔE n +T*E n +D×Δ(ΔE n );

[0159] K = K p ×K s ;

[0160]

[0161]

[0162] K p =100 / PB;

[0163] K s = (MSH-MSL) / (SH-SL);

[0164] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable; when n < 6, Kn, Tn, and Dn are 0; E n ΔE represents the deviation between the measured value and the set value at time n. n E at time n n E at time n-1 n The deviation; Δ(ΔE) n ) represents ΔE at the nth time. n ΔE at time n-1 n Deviation; MV n The output value at time n; ΔMV n MV at time n n MV at time n-1 n The deviation.

[0165] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0166] The gain parameter K, integral time parameter T, and derivative time parameter D of the deviation-based proportional-integral-derivative (PID) control are obtained by solving the deviation-based PID model and the deviation calculation model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; Dn is the calculation formula for the derivative time parameter D at time n; where:

[0167]

[0168]

[0169]

[0170] In the above formula: a1=dEsn-2, a2=dEsn-1, a3=dEsn, b1=Esn-2, b2=Esn-1, b3=Esn, c1=ddEsn-2, c2=ddEsn-1, c3=ddEsn, d1=dMvsn-2, d2=dMvsn-1, d3=dMvsn; n is the time position of the time series.

[0171] In one possible embodiment, the deviation-based proportional-integral (PI) control model is as follows:

[0172] ΔMV n =K×ΔE n +T×En ;

[0173] K = K p ×K s ;

[0174]

[0175] K p =100 / PB;

[0176] K s = (MSH-MSL) / (SH-SL);

[0177] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable.

[0178] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0179] The gain parameter K and integral time parameter T of the deviation-based proportional-integral (PI) control are obtained by solving the deviation-based PI model and the deviation calculation model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; when n < 4, Kn, Tn, and Dn are 0; where:

[0180]

[0181]

[0182] In one possible embodiment, the deviation-based proportional control (P) model is as follows:

[0183] ΔMV n =K×ΔE n ;

[0184] K = K p ×K s ;

[0185] K p =100 / PB;

[0186] K s = (MSH-MSL) / (SH-SL);

[0187] Where Kp is the process gain, PB is the proportional gain, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable; when n < 4, Kn and Tn are 0.

[0188] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0189] The gain parameter K corresponding to the bias-based proportional control (P) model is obtained by solving the bias-based proportional control (P) model and the bias calculation model; Kn is the calculation formula for the gain parameter K at the nth time step; where:

[0190]

[0191] In one possible embodiment, the deviation-based integral control (I) model is as follows:

[0192] ΔMV n =T×E n ;

[0193]

[0194] K p =100 / PB;

[0195] K s = (MSH-MSL) / (SH-SL);

[0196] Where Kp is the process gain, PB is the proportional gain, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable.

[0197] Applying at least one discrete control model to the deviation sequence, the model reference parameters corresponding to each discrete control model are determined, including:

[0198] The integral time parameter T corresponding to the deviation-based integral control (I) model is obtained by solving the deviation-based integral control (I) model and the deviation calculation model; Tn is the calculation formula for the integral time parameter T at the nth time step; where:

[0199] When n < 2, Tn is 0.

[0200] Optionally, the proportional-integral control (PI-D) model based on the derivative of the measured value and the deviation is as follows:

[0201] ΔMV n =K×ΔE n +T×E n+D×Δ(ΔPV n );

[0202]

[0203] K = K p ×K s ;

[0204]

[0205] K p =100 / PB;

[0206] K s = (MSH-MSL) / (SH-SL);

[0207] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable; when n < 6, Kn, Tn, and Dn are 0; PV n The measurement value at time n; ΔPV n For the PV at time n n PV at time n-1 n The deviation; Δ(ΔPV) n ) represents the ΔPV at time n. n ΔPV at time n-1 n Deviation; MV n The output value at time n; ΔMV n MV at time n n MV at time n-1 n The deviation.

[0208] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0209] Solving the proportional-integral-d (PI-D) control model based on the derivative and deviation of the measured values, along with the deviation calculation model, yields the gain parameter K, integral time parameter T, and derivative time parameter D corresponding to the PI-D control based on the derivative and deviation of the measured values; Kn is the calculation formula for the gain parameter K at the nth time; Tn is the calculation formula for the integral time parameter T at the nth time; Dn is the calculation formula for the derivative time parameter D at the nth time; where:

[0210]

[0211]

[0212]

[0213] In the above formula: a1=dEsn-2, a2=dEsn-1, a3=dEsn, b1=Esn-2, b2=Esn-1, b3=Esn, c1=ddPVsn-2, c2=ddPVsn-1, c3=ddPVsn, d1=dMvsn-2, d2=dMvsn-1, d3=dMvsn; n is the time position of the time series; when n<6, Kn, Tn, and Dn are 0.

[0214] In one possible embodiment, the proportional-derivative model based on the measured values ​​and the integral-PD model based on the deviation are as follows:

[0215] ΔMV n =K×ΔPV n +T×E n +D×Δ(ΔPV n );

[0216]

[0217] K = K p ×K s ;

[0218]

[0219] K p =100 / PB;

[0220] K s = (MSH-MSL) / (SH-SL);

[0221] Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, MSL is the lower limit of the manipulated variable; PV n The measurement value at time n; ΔPV n PV at time n n PV at time n-1 n The deviation; Δ(ΔPV) n ) represents the ΔPV at time n. n ΔPV at time n-1 n Deviation; MV n The output value at time n; ΔMV n MV at time n n MV at time n-1 n The deviation.

[0222] Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including:

[0223] Solving the measured value-based proportional-derivative model, the deviation integral control (I-PD) model, and the deviation calculation model yields the gain parameter K, integral time parameter T, and derivative time parameter D corresponding to the measured value-based proportional-derivative model and the deviation integral control (I-PD) model; Kn is the calculation formula for the gain parameter K at the nth time; Tn is the calculation formula for the integral time parameter T at the nth time; Dn is the calculation formula for the derivative time parameter D at the nth time; where:

[0224]

[0225]

[0226]

[0227] In the above formula: a1=dPVsn-2, a2=dPVsn-1, a3=dPVsn, b1=Esn-2, b2=Esn-1, b3=Esn, c1=ddPVsn-2, c2=ddPVsn-1, c3=ddPVsn, d1=dMvsn-2, d2=dMvsn-1, d3=dMvsn; n is the time position of the time series; when n<6, Kn, Tn, and Dn are 0.

[0228] Optionally, the step of using a parameter identification probability evaluation model to perform identification probability analysis on all reference parameter sequences to obtain the identification rate and identification value corresponding to all reference parameter sequences, wherein the identification rate and identification value of each reference parameter sequence correspond to each other, including:

[0229] S1) Determine the number of equal-value segments SPCount in the value range of the identification sequence B; (e.g., the number of segments from the minimum to the maximum value of the B sequence, for example, values ​​1-10, according to 9 segments, with a segmentation interval of 1, are decomposed into 1~2, 2~3, 3~4, 4~5...) wherein the identification sequence B is the reference parameter sequence;

[0230] S2) Determine the number of data items nCount in the subset newB of the identification sequence B; the data item size of the subset newB is equal to the data item size of the identification sequence B;

[0231] S3) Determine the data segmentation interval sp based on the maximum and minimum values ​​in the subset newB and the number of segments SPCount;

[0232] S4) Based on the number of segments SPCount and the data segmentation interval sp, initialize the segmentation interval sequence sps and the amount of data within the data segmentation interval sp;

[0233] S5) Based on the value of subset newB, determine the number of data points spcount that fall within the range of each data partition interval sp; (e.g., the number of data points in partition intervals 1-2, 2-3, etc.).

[0234] S6) Obtain the maximum value spmax from the number of data items spcount;

[0235] S7) Remove data from other segmentation periods whose data volume is less than a preset value (other segmentation periods are between spcount-sp / 2 and spcount+sp / 2), store the removed subset newB and update the data volume of the subset newB; the preset value is a preset multiple of the maximum value spmax; preferably, the preset multiple of the maximum value spmax is preferably 0.01 times the maximum value spmax.

[0236] S8) Repeat steps S2)-S7) until the number of iterations (preferably 10) reaches the set value or the data size of subset newB is less than the number of partitions SPCount. Then, output the data size of subset newB as the data size of the result subset newB and stop the loop.

[0237] S9) Obtain the average data size avg of the result subset newB, and select the subset newB whose data size and average avg satisfy a preset condition; the preset condition is that the data size of newB should not be less than the data size of the original identification sequence B. That is, it is necessary to remove data from newB whose data size is less than 0.5avg or whose data size is greater than 0.5avg, and select one of them to remove data according to the requirements;

[0238] S10) Obtain the average value avg of all selected subsets newB. Calculate the similarity S between the selected subset newB and its average value avg. The similarity S is calculated using the following formula:

[0239] S n =avg / newB n ,(avg>0)and(newB n ≥avg)

[0240] S n =newB n / avg, (avg>0)and(newBn <avg)

[0241] S n =newB n / avg, (avg < 0) and (newB) n ≥avg)

[0242] S n =avg / newB n ,(avgg<0)and(newB n <avg)

[0243] Sn represents the similarity S value at time n. Then, the average value of the obtained similarity S values ​​is calculated to obtain the average value (Savg) of the similarity S values ​​of the selected subset newB relative to the selected subset newB.

[0244] S11) Based on the average value Savg of the similarity degree S, the value of the selected subset newB, and the amount of data in the selected subset newB, calculate the recognition rate and recognition value corresponding to the amount of data in the selected subset newB, and use them as the recognition rate and recognition value corresponding to the reference parameter sequence, respectively.

[0245] Optionally, the recognition rate and recognition value corresponding to the data volume of the selected subset newB can be calculated using the following formula:

[0246] B rely =Savg × nCount / AllCount.

[0247] B avg =∑newB n / nCount;

[0248] Among them, B rely For recognition rate; B avg For identification value; newB n The data size of the selected subset newB is ; AIICount is the number of data items in the identification sequence B.

[0249] Furthermore, if the recognition rate of all reference parameter sequences corresponding to the model reference parameter set is greater than a preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop, and the nonlinearity of the control loop is determined based on the maximum recognition rate among all recognition rates corresponding to the discrete control model of the control loop; the specific process is as follows:

[0250] Calculate the K, T, D identification values ​​and identification rate arrays (K, T, D) of various PID control models (PID, PI-D, I-PD, PI, P, I) based on the PID parameter identification method. PID T PIDD PID (K) PI T PI (K) PI ), (T PI (K) PI-D T PI-D D PI-D (K) I-PD T I-PD D I-PD If the recognition rate of each parameter is greater than 35%, the recognition of that parameter is considered valid. The recognition rates and values ​​of PID, PI-D, I-PD, PI, P, and I of the control loop acquisition and calculation of the regulating valve FIC_20301 in this embodiment are shown in Table 1 below:

[0251] Table 1

[0252]

[0253] According to the PID control model discrimination rules, the PID control model of FIC_20301 can be identified as a PI model, with a discrimination value of K of -0.0011 and a discrimination rate of 96.7%, and a discrimination value of T of 0.00000028 and a discrimination rate of 54.0%. The integral time of this loop is too short, essentially operating under ratio control. Therefore, the degree of nonlinearity corresponding to the regulating valve FIC_20301 is:

[0254] The degree of nonlinearity of FIC_20301 = 100 - K rely =3.3%.

[0255] The above-mentioned linear operating probability time series of the control loop is determined based on the reference parameter sequence corresponding to the maximum recognition rate and the maximum recognition rate.

[0256] The maximum recognition rate K was obtained based on the recognition rates of the K and T values ​​in FIC_20301. rely =96.7%, the recognition value K at the maximum recognition rate avg And the identification sequence K.

[0257] Based on the recognition sequence K with the maximum recognition rate of FIC_20301, and the recognition value K... avg The similarity score is 0.0011 for evaluating the linear operating probability time series Sn of the control loop FIC_20301, where:

[0258] S n =K avg / K n , (K avg >0)and(K n ≥K avg )

[0259] S n =K n / K avg , (K avg >0)and(K n <K avg )

[0260] If, S n If S > -200, then let S n =-200.

[0261] The identification rates and identification values ​​of PID, PI-D, I-PD, PI, P, and I of FIC_20301, which are used for control loop acquisition and calculation of the second regulating valve LIC_20702 in this embodiment, are shown in Table 2 below:

[0262] Table 2

[0263]

[0264] According to the PID control model discrimination rules, the PID control model of LIC_20702 can be identified as a PI model, with a discrimination value of K of -0.8 and a discrimination rate of 73.2%, and a discrimination value of T of -0.0013 and a discrimination rate of 54.0%. Negative discrimination values ​​indicate a reaction effect from the controller. The degree of nonlinearity corresponding to the regulating valve LIC_20702 is:

[0265] The nonlinearity of LIC_20702 is 100-K. rely =26.8%.

[0266] The maximum recognition rate K was obtained based on the recognition rates of the K and T values ​​of LIC_20702. rely =73.2%, the recognition value K at the maximum recognition rate avg And the identification sequence K.

[0267] Based on the recognition sequence K with the maximum recognition rate of LIC_20702, and the recognition value K avg = -0.8 is used to evaluate the linear running probability time series Sn of the control loop LIC_20702, where:

[0268] S n =Kn / K avg , (K avg <0)and(K n ≥K avg )

[0269] S n =K avg / K n , (K avg <0)and(K n <Kavg )

[0270] If, S n If S > -200, then let S n =-200.

[0271] Therefore, the linear operating probability parameter set sequence of the control loop is determined as a linear sequence based on the types of any two consecutive markers. Specifically, in one embodiment, based on the FIC_20301 linear operating probability time series S, the time series S is divided into groups of five data points each, S5. If any of the five data points in group S5 has an identification value less than 90%, it is marked as nonlinear (1); otherwise, it is marked as linear (0). Starting from the first group of data in group S5, the linear sequence group and its length are identified. If two consecutive nonlinear sequences appear, the linear sequence segment ends; otherwise, the segment is a linear sequence. It can be concluded that the number of nonlinearities in FIC_20301 is relatively small, indicating that it is essentially in a linear operating state.

[0272] In one embodiment, based on the LIC_20702 linear running probability time series S, linearly stable segments are identified, and the longest sequence is taken as the linear sequence. The time series S is grouped into sequences S5 of every five data points; if any of the five data points in a group S5 has an identification value less than 90%, it is marked as non-linear (1); otherwise, it is marked as linear (0). Starting from the first group of data in group S5, linear sequence groups and their lengths are identified. If two consecutive non-linear sequences appear, the linear sequence segment ends; otherwise, the data segment is a linear sequence. Figure 4 As shown, through existing computer simulation algorithms, it can be concluded that LIC_20702 has a more non-linear distribution from 2 PM to 5 PM on July 1st, and is basically in a linear operating state from 5 PM to 6 PM on July 1st.

[0273] This patent uses mathematical statistical methods to identify PID parameters. Based on the identification probability of PID parameters, it evaluates the nonlinear disturbance at each sampling time in the closed-loop control process and the time series of nonlinear disturbances, diagnoses the nonlinear sequences, and filters the nonlinear sequences, which is beneficial for linear control modeling and valve fault diagnosis.

[0274] The present invention also provides a machine-readable storage medium storing instructions for causing a machine to execute the above-described closed-loop nonlinear evaluation method based on PID parameter identification.

[0275] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0276] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0277] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A closed-loop nonlinearity evaluation method based on PID parameter identification, characterized in that, The nonlinear evaluation method includes: The operating data of the control loop of the refining and chemical unit at multiple sampling times are obtained, and the deviation data corresponding to the sampling times are obtained based on the operating data at the multiple sampling times and a deviation sequence is formed according to the order of the sampling times. The deviation sequence is applied to at least one discrete control model to obtain a model reference parameter set corresponding to each discrete control model. Each reference parameter in the model reference parameter set forms a reference parameter sequence according to the order of sampling time. The identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate of all reference parameter sequences; If the identification rate of all reference parameter sequences corresponding to the model reference parameter set is greater than the preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop, and the nonlinearity of the control loop is determined according to the maximum identification rate among all identification rates corresponding to the discrete control model of the control loop. The discrete control model includes one or more of the following: deviation-based proportional-integral-derivative (PID) control model, deviation-based proportional-integral (PI) control model, deviation-based proportional control (P) model, deviation-based integral control (I) model, deviation-based proportional-integral (PI-D) control model based on the derivative of the measured value and the deviation, and deviation-based proportional-derivative and deviation-integral (I-PD) control model. This includes acquiring operational data of the refining unit's control loop at multiple sampling times, obtaining deviation data corresponding to the sampling times based on the operational data at these multiple sampling times, and forming a deviation sequence according to the chronological order of the sampling times, including: Acquire the setpoint SP, measured value PV, and output value MV of the refining unit control loop at multiple sampling times; The following deviation calculation model is used to calculate the deviation data corresponding to the sampling time and form a deviation sequence: ； Among them, PV n SP is the PV value at time n; n E is the SP value at time n; n Let E be the deviation E between the measured value PV and the set value SP at time n; dE n Let dE be the deviation value of the deviation E at time n. When n is less than 2, dE is... n =0; dPV n dPV is the deviation of the PV value at time n, where n is less than 2. n =0; dMV n dMV is the deviation of the MV value at time n, where n is less than 2. n =0;ddE n ddE is the deviation of dE at time n. When n is less than 3, ddE n =0; ddPV n The deviation of dPV at time n is the deviation value when n is less than 3. n =0.

2. The nonlinear evaluation method according to claim 1, characterized in that, The nonlinear evaluation method further includes: The identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate and identification value corresponding to each reference parameter sequence; the identification rate and identification value of each reference parameter sequence correspond to each other; If the identification rate of all reference parameter sequences corresponding to the model reference parameter set is greater than the preset value, the discrete control model corresponding to the model reference parameter set is determined to be the discrete control model of the control loop. The maximum identification rate and the identification value corresponding to the maximum identification rate are determined among all the identification rates of the discrete control model of the control loop, and the reference parameter sequence corresponding to the maximum identification rate is determined. Based on the reference parameter sequence corresponding to the maximum recognition rate and the maximum recognition rate, the linear operating probability time series of the control loop is determined.

3. The nonlinear evaluation method according to claim 2, characterized in that, The nonlinear evaluation method further includes: The linear operating probability time series of the control loop is continuously grouped according to a preset number of intervals to obtain a continuously grouped sequence; The marker for each sequence group is determined by the magnitude of the identification value corresponding to each group in the continuous grouping sequence; The linear operating probability parameter set sequence of the control loop is determined based on the types of any two consecutive markers.

4. The nonlinear evaluation method according to claim 3, characterized in that, The markers include non-linear markers and linear markers; The markers used to determine each sequence group include: If any identification value in the corresponding sequence group is less than the second preset value, the sequence group is marked with a non-linear marker; otherwise, the sequence group is marked with a linear marker. The step of determining whether the sequence of linear operating probability parameters of the control loop is a linear sequence based on the types of any two consecutive markers includes: If any two consecutive sets of sequences have markers of the type of non-linear markers and linear markers respectively, or if any two consecutive sets of sequences have markers of the type of linear markers and non-linear markers respectively, then the two consecutive sets of sequences are determined to be linear sequences.

5. The nonlinear evaluation method according to claim 2, characterized in that, The parameter identification probability evaluation model is used to perform identification probability analysis on all reference parameter sequences to obtain the identification rate and identification value corresponding to each reference parameter sequence. The identification rate and identification value of each reference parameter sequence correspond to each other, including: S1) Determine the number of equal-value segments SPCount in the value range of the identification sequence B; wherein the identification sequence B is a reference parameter sequence; S2) Determine the number of data items nCount in the subset newB of the identification sequence B; the number of data items in the subset newB is equal to the number of data items in the identification sequence B. S3) Determine the data segmentation interval sp based on the maximum and minimum values ​​in the subset newB and the number of segments SPCount; S4) Based on the number of segments SPCount and the data segmentation interval sp, initialize the segmentation interval sequence sps and the amount of data within the data segmentation interval sp; S5) Based on the value of subset newB, determine the number of data items spcount whose values ​​conform to the range of each data partition interval sp; S6) Obtain the maximum value spmax from the number of data items spcount; S7) Remove data from other segmentation periods whose data volume is less than a preset value, store the removed subset newB and update the data volume of the subset newB; the preset value is a preset multiple of the maximum value spmax. S8) Repeat steps S2)-S7) until the number of iterations reaches the set value or the data size of subset newB is less than the number of splits SPCount, then output the data size of subset newB as the data size of the result subset newB and stop the loop; S9) Obtain the average value avg of the data volume of the result subset newB, and select the subset newB whose data volume value and average value avg satisfy the preset condition as the selected subset newB; S10) Obtain the average value avg of all selected subsets newB. Calculate the similarity S between the value of selected subset newB and the average value avg of selected subset newB, and obtain the average value Savg of the similarity S between the value of selected subset newB and the average value avg of selected subset newB. S11) Based on the average value Savg of the similarity degree S, the value of the selected subset newB, and the amount of data in the selected subset newB, calculate the recognition rate and recognition value corresponding to the amount of data in the selected subset newB, and use them as the recognition rate and recognition value corresponding to the reference parameter sequence, respectively.

6. The nonlinear evaluation method according to claim 5, characterized in that, The recognition rate and recognition value corresponding to the data volume of the selected subset newB are calculated using the following formulas: ； ； Among them, B rely For recognition rate; B avg For identification value; newB n The value of the nth selected subset newB; AllCount is the number of data items in the identification sequence B.

7. The nonlinear evaluation method according to any one of claims 1-6, characterized in that, The model reference parameter set includes one or more of the following: gain parameter K, integral time parameter T, and derivative time parameter D.

8. The nonlinear evaluation method according to claim 1, characterized in that, The deviation-based proportional-integral-derivative (PID) control model is as follows: ； ； ； ； ； ； Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling coefficient, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable. This represents the deviation between the measured value and the set value at the nth time point; For the nth time With the (n-1)th time Deviation; For the nth time With the (n-1)th time Deviation; This is the output value at time n; For the nth time With the (n-1)th time Deviation; Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including: The gain parameter K, integral time parameter T, and derivative time parameter D of the deviation-based proportional-integral-derivative (PID) control are obtained by solving the deviation-based PID model and the deviation calculation model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; Dn is the calculation formula for the derivative time parameter D at time n; where: ; ; ; In the above formula: a1=dEn-2, a2=dEn-1, a3=dEn, b1=En-2, b2=En-1, b3=En, c1=ddEn-2, c2=ddEn-1, c3=ddEn, d1=dMvn-2, d2=dMvn-1, d3=dMvn; n is the time position of the time series.

9. The nonlinear evaluation method according to claim 1, characterized in that, The proportional-integral (PI) control model based on deviation is as follows: ； ； ； ； ； Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable. This represents the deviation between the measured value and the set value at the nth time point; For the nth time With the (n-1)th time Deviation; This is the output value at time n; For the nth time With the (n-1)th time Deviation; Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including: The gain parameter K and integral time parameter T of the deviation-based proportional-integral (PI) control are obtained by solving the deviation-based PI model and the deviation calculation model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; where: ； 。 10. The nonlinear evaluation method according to claim 1, characterized in that, The proportional control (P) model based on deviation is as follows: ； ； ； ； Where Kp is the process gain, PB is the proportional gain, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable. This represents the deviation between the measured value and the set value at the nth time point; For the nth time With the (n-1)th time Deviation; This is the output value at time n; For the nth time With the (n-1)th time Deviation; Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including: The gain parameter K corresponding to the bias-based proportional control (P) model is obtained by solving the bias-based proportional control (P) model and the bias calculation model; Kn is the calculation formula for the gain parameter K at the nth time step; where: 。 11. The nonlinear evaluation method according to claim 1, characterized in that, The deviation-based integral control (I) model is as follows: ； ； ； ； Where Kp is the process gain, PB is the proportional gain, Ks is the scaling factor, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable. This represents the deviation between the measured value and the set value at the nth time point; This is the output value at time n; For the nth time With the (n-1)th time The deviation, where TI is the integration time; Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including: Solving the deviation-based integral control (I) model and the deviation calculation model yields the integral time parameter T corresponding to the deviation-based integral control (I); Tn is the calculation formula for the integral time parameter T at the nth time step; where: 。 12. The nonlinear evaluation method according to claim 1, characterized in that, The proportional-integral control (PI-D) model based on the derivative of the measured value and the deviation is as follows: ； ； ； ； ； ； Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling coefficient, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable. This is the measurement value at the nth time point; For the nth time With the (n-1)th time Deviation; For the nth time With the (n-1)th time Deviation; This is the output value at time n; For the nth time With the (n-1)th time Deviation; Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including: Solving the proportional-integral-d (PI-D) control model based on the derivative and deviation of the measured values, along with the deviation calculation model, yields the gain parameter K, integral time parameter T, and derivative time parameter D corresponding to the PI-D control based on the derivative and deviation of the measured values; Kn is the calculation formula for the gain parameter K at the nth time; Tn is the calculation formula for the integral time parameter T at the nth time; Dn is the calculation formula for the derivative time parameter D at the nth time; where: ; ; ； In the above formula: a1=dEn-2, a2=dEn-1, a3=dEn, b1=En-2, b2=En-1, b3=En, c1=ddPVn-2, c2=ddPVn-1, c3=ddPVn, d1=dMvn-2, d2=dMvn-1, d3=dMvn; n is the time position of the time series.

13. The nonlinear evaluation method according to claim 1, characterized in that, The proportional-derivative model based on measured values ​​and the integral-PD model based on deviation are as follows: ； ； ； ； ； ； Where ΔT is the control period, Kp is the process gain, PB is the proportional gain, TI is the integral time, TD is the derivative time, Ks is the scaling coefficient, SH is the upper limit of the process variable, SL is the lower limit of the process variable, MSH is the upper limit of the manipulated variable, and MSL is the lower limit of the manipulated variable. This is the measurement value at the nth time point; For the nth time With the (n-1)th time Deviation; For the nth time With the (n-1)th time Deviation; This is the output value at time n; For the nth time With the (n-1)th time Deviation; Applying at least one discrete control model to the deviation sequence, a model reference parameter set corresponding to each discrete control model is obtained, including: Solving the measured value-based proportional-derivative model, the deviation integral control (I-PD) model, and the deviation calculation model yields the gain parameter K, integral time parameter T, and derivative time parameter D corresponding to the measured value-based proportional-derivative model and the deviation integral control (I-PD) model; Kn is the calculation formula for the gain parameter K at time n; Tn is the calculation formula for the integral time parameter T at time n; Dn is the calculation formula for the derivative time parameter D at time n; where: ; ; ； In the above formula: a1=dPVn-2, a2=dPVn-1, a3=dPVn, b1=En-2, b2=En-1, b3=En, c1=ddPVn-2, c2=ddPVn-1, c3=ddPVn, d1=dMvn-2, d2=dMvn-1, d3=dMvn; n is the time position of the time series.

14. A machine-readable storage medium storing instructions for causing a machine to perform the closed-loop nonlinear evaluation method based on PID parameter identification as described in any one of claims 1-13.

Images