Internal model control system based on real-time adaptive online identification of industrial system

Abstract

The invention relates to the technical field of data processing of a control system, in particular to an internal model control system based on real-time self-adaptive online identification of an industrial system, which comprises a real-time data acquisition unit for acquiring an actuator opening instruction value, an actual controlled system output value and a model controlled system output value; the deviation calculation unit outputs a deviation value between the actual output value of the controlled system and the output value of the model controlled system; the dynamic parameter identification unit generates a weight factor in a linear proportional relationship with the square of the deviation value, outputs an adjusted gain parameter when the weight factor is greater than a preset threshold value, and outputs an adjusted time constant parameter when the weight factor is smaller than the preset threshold value; the internal model control model updating unit reconstructs a transfer function model; the controller parameter optimization unit outputs an optimized actuator opening instruction value; and the control execution unit drives the execution mechanism to enable the actual controlled system output value to respond to the instruction change. According to the method, the problems of dynamic response degradation and steady-state error increase caused by fixed parameter model mismatch are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of control system data processing, and in particular to an internal model control system based on real-time adaptive online identification of an industrial system. BACKGROUND

[0002] Industrial systems include DCS systems, PLC systems, SCADA systems and the like, and the following technical pain points exist in the online identification technology in existing industrial systems: the existing adaptive scheme is limited by the industrial control platform architecture: an external device scheme: can support non-real-time parameter updating, but has the problems of high hardware cost and data synchronization delay, a DCS native scheme: is limited by strict time domain constraints, and the existing adaptive algorithm cannot meet the real-time requirements, specifically, the time-varying characteristics of the industrial process are not compensated in real time, for example, in the boiler desuperheating water control system of a power plant, the high-temperature and high-pressure environment continuously acts to cause the mechanical characteristics of the regulating valve to gradually change, the response rate and flow capacity of the actuator to irreversibly drift, and the gain and time constant of the transfer function of the controlled object to significantly deviate; while the internal model control architecture still maintains the fixed parameter model in the initial debugging stage, the model output cannot track the changes in the actual process output, the identification deviation continuously accumulates, causing the actuator opening degree command to lag and oscillate, the main steam temperature regulation process to have excessive overshoot, and the temperature value to continuously deviate from the process set range after long-term operation, affecting the thermal efficiency of the unit and the service life of the equipment. SUMMARY

[0003] In view of the deficiencies of the prior art, the present application provides an internal model control system based on real-time adaptive online identification of an industrial system, which solves the technical problem of parameter mismatch of the fixed parameter model of the internal model control, and further causes dynamic response degradation and steady-state error increase, caused by the drift of the transfer function of the controlled object due to the time-varying characteristics of the industrial process, such as equipment aging or load fluctuation.

[0004] To solve the above technical problems, the present application is as follows: The internal model control system based on real-time adaptive online identification of an industrial system provided by the present application comprises: a real-time data acquisition unit for acquiring an actuator opening degree command value, an actual controlled system output value and a model controlled system output value; a deviation calculation unit for receiving the actual controlled system output value and the model controlled system output value output by the real-time data acquisition unit and outputting a deviation value of the actual controlled system output value and the model controlled system output value; a dynamic parameter identification unit for receiving the deviation value output by the deviation calculation unit, generating a weight factor in a linear proportional relationship with the square of the deviation value, outputting an adjusted gain parameter when the weight factor is greater than a preset threshold value, and outputting an adjusted time constant parameter when the weight factor is less than the preset threshold value; An inner model control model updating unit receives the gain parameter and the time constant parameter output by the dynamic parameter identification unit, and outputs a reconstructed transfer function model; A controller parameter optimization unit receives the actuator opening degree instruction value acquired by the real-time data acquisition unit, the target set value, and the transfer function model output by the inner model control model updating unit, and outputs an optimized actuator opening degree instruction value. A control execution unit receives the optimized actuator opening degree instruction value output by the controller parameter optimization unit, and outputs a driving signal to an actuator to make the actual controlled system output value change in response to the optimized actuator opening degree instruction value.

[0005] Further, the inner model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The real-time data acquisition unit, the deviation calculation unit, and the dynamic parameter identification unit are implemented through a function block of a distributed control system (DCS), and the parameter identification period is less than or equal to 200 ms, and the steam temperature data is acquired from a boiler temperature reducing water system of a power plant.

[0006] Further, the inner model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The deviation calculation unit receives the steam temperature data output by the real-time data acquisition unit, calculates an instantaneous deviation value of a temperature set point from an actual measured value in a denitration control system through a subtraction function block, and outputs the instantaneous deviation value to the dynamic parameter identification unit.

[0007] Further, the inner model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The dynamic parameter identification unit receives the instantaneous deviation value, generates a weight factor greater than 0.8 to output an adjusted gain parameter in an initial stage, and generates a weight factor less than 0.3 to output an adjusted time constant parameter in a subsequent stage.

[0008] Further, the inner model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The inner model control model updating unit receives the adjusted gain parameter and the time constant parameter, reconstructs a transfer function model through a variable parameter transfer function function block, and updates parameters in response to load fluctuations in a circulating fluidized bed unit main steam pressure control.

[0009] Further, the inner model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The controller parameter optimization unit receives an updated transfer function model and a target set value in the main gas pressure control of the circulating fluidized bed unit, and outputs a continuously adjusted actuator opening degree instruction value, and outputs a continuously adjusted actuator opening degree instruction value to the control execution unit in the shaft seal temperature reduction control, so that the actual controlled system output value responds to the instruction change.

[0010] Further, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The control execution unit receives the actuator opening degree instruction value output in the shaft seal temperature reduction control, and drives the regulating valve in the supercritical boiler main control system, so that the error of the actual output value of the steam pressure tracking the target set value approaches zero.

[0011] Further, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: A bidirectional decoupling unit is arranged between the controller parameter optimization unit and the control execution unit, and the bidirectional decoupling unit performs: Receiving the actuator opening degree instruction value output by the first controller parameter optimization unit and the actuator opening degree instruction value output by the second controller parameter optimization unit; Generating the decoupled first actuator opening degree instruction value and the second actuator opening degree instruction value through cross parameter compensation; Outputting the decoupled first actuator opening degree instruction value to the first control execution unit and the second actuator opening degree instruction value to the second control execution unit.

[0012] Further, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application, the bidirectional decoupling unit is configured to: Identifying the first gain parameter and the first time constant parameter from the actual controlled system output value fed back by the first control execution unit; Identifying the second gain parameter and the second time constant parameter from the actual controlled system output value fed back by the second control execution unit; Inputting the first gain parameter and the first time constant parameter into the first internal model control model updating unit; Inputting the second gain parameter and the second time constant parameter into the second internal model control model updating unit.

[0013] Further, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The bidirectional decoupling unit updates the first gain parameter, the first time constant parameter, the second gain parameter and the second time constant parameter in real time in response to load fluctuation in the main steam pressure and primary air pressure double control system of the power plant circulating fluidized bed unit.

[0014] The present application has the following beneficial effects: The application creatively solves the adaptation bottleneck of the industrial control platform: the DCS originality, for the first time realizes 200ms level parameter real-time identification in the distributed control system native environment, reduces the delay compared with the external scheme, Cross-platform value, the same algorithm is also applicable in PLC / SCADA and other external devices, but the DCS implementation significantly improves the system integration, industrial universality, covers boiler desuperheating water control (steam temperature), denitration control (NOx concentration) and other multiple scenes, solves the model mismatch problem caused by the time-varying characteristics of the industrial process: the real-time data acquisition unit obtains the actuator opening command value, the actual controlled system output value and the model controlled system output value, and constructs the closed-loop control initial data source; the deviation calculation unit outputs the deviation value of the actual controlled system output value and the model controlled system output value, quantifies the instantaneous error caused by device aging or load fluctuation; the dynamic parameter identification unit generates a weight factor in a linear proportional relationship with the square of the deviation value, and when the weight factor is greater than the preset threshold, the adjusted gain parameter is output to compensate for the valve characteristic drift, and when the weight factor is less than the preset threshold, the adjusted time constant parameter is output to optimize the dynamic response speed; the internal model control model updating unit receives the gain parameter and the time constant parameter to reconstruct the transfer function model, and updates the parameters in response to load fluctuations in the circulating fluidized bed unit main steam pressure control; the controller parameter optimization unit fuses the target set value and the updated transfer function model to output the continuously adjusted actuator opening command value, eliminating the overshoot phenomenon in the shaft seal desuperheating control; the control execution unit drives the regulating valve to make the error between the actual output value of the steam pressure and the target set value approach zero; the bidirectional decoupling unit generates a decoupling instruction value through cross-parameter compensation, updates the grouped gain parameter and time constant parameter in the power plant circulating fluidized bed main steam pressure and primary air pressure double control, and eliminates the bidirectional coupling interference; the application forms a complete logical chain of "deviation awareness, weight factor staged parameter adjustment, model dynamic reconstruction and decoupling control expansion", breaking through the dynamic response degradation and steady-state error increase bottleneck caused by fixed parameter model mismatch. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the application of the application, the drawings needed in the embodiments will be briefly introduced below, and obviously, other drawings can also be obtained by the drawings for those skilled in the art without creative labor.

[0016] Figure 1 The system architecture diagram of the internal model control system based on industrial system real-time self-adaptive online identification provided for the embodiments of the application.

[0017] Figure 2 The DCS online identification principle diagram provided for the embodiments of the application.

[0018] Figure 3The online identification dynamic response diagram provided by the embodiment of the present application.

[0019] Figure 4 The IMC integrated control architecture diagram provided by the embodiment of the present application.

[0020] Figure 5 The decoupling control effect comparison diagram provided by the embodiment of the present application.

[0021] Figure 6 The MIMO bidirectional decoupling structure diagram provided by the embodiment of the present application.

[0022] Figure 7 The MIMO system identification effect diagram provided by the embodiment of the present application.

[0023] Figure 8 The MIMO online identification coordination architecture diagram provided by the embodiment of the present application.

[0024] Figure 9 The control optimization result diagram provided by the embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to make the purpose of the present application, the present application and the advantages clearer, the present application will be described clearly and completely below by combining the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only one module embodiment of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts belong to the protection scope of the present application. The present application provided by each embodiment of the present application will be described in detail below by combining the drawings. In order to better understand the purpose of the present application, the present application will be further described in detail below.

[0026] Please refer to Figures 1 to 9 The internal model control system based on real-time adaptive online identification of an industrial system provided by the present application comprises: The present application realizes the decoupling of the algorithm and the hardware platform through a weight factor linear calculation model. The 200ms period identification is realized through the function block configuration on the DCS platform. When the external control device (such as PLC, SCADA system) is used, the same algorithm library is called through the OPC data interface, and the period can be configured to 1s-10s.

[0027] The real-time data acquisition unit acquires the actuator opening degree instruction value, the actual controlled system output value and the model controlled system output value; The deviation calculation unit receives the actual controlled system output value and the model controlled system output value output by the real-time data acquisition unit, and outputs the deviation value of the actual controlled system output value and the model controlled system output value; The dynamic parameter identification unit receives the deviation value output by the deviation calculation unit, generates a weight factor in a linear proportional relationship with the square of the deviation value, outputs an adjusted gain parameter when the weight factor is greater than a preset threshold, and outputs an adjusted time constant parameter when the weight factor is less than the preset threshold. The internal model control model updating unit receives the gain parameter and the time constant parameter output by the dynamic parameter identification unit, and outputs a reconstructed transfer function model. The controller parameter optimization unit receives the actuator opening degree instruction value, the target set value, and the transfer function model output by the internal model control model updating unit, and outputs an optimized actuator opening degree instruction value. The control execution unit receives the optimized actuator opening degree instruction value output by the controller parameter optimization unit, and outputs a driving signal to the actuator to make the actual controlled system output value respond to the optimized actuator opening degree instruction value.

[0028] The real-time data acquisition unit synchronously acquires the actuator opening degree instruction value, the actual controlled system output value, and the model controlled system output value from the industrial field distributed control system. The actuator opening degree instruction value represents the opening control signal of the regulating valve, the actual controlled system output value reflects the real-time state variable of the industrial process, and the model controlled system output value is the predicted value of the internal model control architecture. In the power plant boiler desuperheating water control scene, the acquisition unit captures the steam temperature data at a fixed frequency through the analog input function block of the distributed control system, and the three types of data are transmitted to the deviation calculation unit through the parallel bus to build the initial data source of the control feedback.

[0029] The deviation calculation unit receives the actual controlled system output value and the model controlled system output value output by the real-time data acquisition unit, and calculates the deviation value of the two by using a subtraction operation function block. The deviation value quantifies the matching error between the model predicted value and the actual process, for example, the instantaneous difference between the temperature set point and the actual measured value in the denitration control system. The deviation value is used as the input of the dynamic parameter identification unit to drive the subsequent parameter adjustment process. Before outputting the deviation value, the unit performs data normalization processing to eliminate the dimensional difference and realize the consistency of the processing of the downstream modules.

[0030] The dynamic parameter identification unit receives the deviation value output by the deviation calculation unit, and generates a weight factor in a linear proportional relationship with the square of the deviation value. The weight factor controls the priority of parameter adjustment: in the initial stage, when the weight factor is greater than a preset threshold, the unit only outputs the adjusted gain parameter to accelerate the system response; in the subsequent stage, when the weight factor is less than the preset threshold, the unit outputs the adjusted time constant parameter to improve the steady-state accuracy. In the main steam pressure control of the circulating fluidized bed unit, the gain parameter compensates for the valve characteristic drift caused by load fluctuation, and the time constant parameter optimizes the dynamic response speed. Both are output to the internal model control model updating unit in stages.

[0031] The internal model control model updating unit receives the gain parameter and the time constant parameter output by the dynamic parameter identification unit, and reconstructs the transfer function model. The reconstruction process adopts a variable parameter transfer function function block, inputs the updated gain parameter and the time constant parameter to generate a dynamic model. The model output is compared with the actual controlled system output value in real time, for example, when responding to the load fluctuation of the circulating fluidized bed unit, the model eliminates the transfer function drift, and outputs the updated transfer function model to the controller parameter optimization unit, to support the real-time optimization of the control algorithm.

[0032] The controller parameter optimization unit receives the actuator opening degree instruction value, the target set value and the transfer function model output by the internal model control model updating unit, and fuses the three types of inputs to generate an optimized actuator opening degree instruction value. In the shaft seal temperature reduction control, the unit performs dynamic weighted calculation based on the updated model, and outputs a continuously adjusted actuator opening degree instruction value to eliminate temperature control overshoot and oscillation. The optimized instruction value is transmitted to the control execution unit through a digital output channel to complete the control strategy closed loop.

[0033] The control execution unit receives the optimized actuator opening degree instruction value output by the controller parameter optimization unit, and outputs a driving signal to the actuator. The driving signal controls the valve opening degree through the output function block of the distributed control system, for example, in the supercritical boiler master control system, the driving valve makes the actual output value of the steam pressure track the target set value, and the dynamic response is stable and the steady-state error tends to zero. The execution result is fed back to the industrial process to form a closed loop control flow of "instruction, execution and output response".

[0034] Specifically, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The real-time data acquisition unit, the deviation calculation unit and the dynamic parameter identification unit are realized by the function block of the distributed control system (DCS), and the parameter identification period is ≤200ms, and the steam temperature data is obtained from the boiler desuperheating water system of the power plant.

[0035] The real-time data acquisition unit acquires data through the analog input function block of the distributed control system, and outputs the steam temperature data to the deviation calculation unit at a frequency of once per second. The steam temperature data comes from the boiler desuperheating water system of the power plant, and the data stream includes three types of synchronous signals, namely the actuator opening degree instruction value, the actual controlled system output value and the model controlled system output value. The high-frequency acquisition of once per second ensures the real-time capture of temperature fluctuations in the boiler desuperheating water control, and the output data packet is transmitted to the downstream deviation calculation unit through the industrial bus to establish the initial data source of the control feedback.

[0036] Specifically, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The deviation calculation unit receives the steam temperature data output by the real-time data acquisition unit, calculates the instantaneous deviation value of the temperature set point and the actual measured value in the denitration control system by using a subtraction function block, and outputs the instantaneous deviation value to the dynamic parameter identification unit.

[0037] The deviation calculation unit receives the process data (such as steam temperature, NOx concentration, etc.) output by the real-time data acquisition unit, calculates the instantaneous deviation value of the set point and the actual measured value by using a subtraction function block, and the process data includes the steam temperature data in the boiler desuperheating water system or the NOx concentration data in the denitration control system. The subtraction operation eliminates the dimension difference and normalizes the process, and the instantaneous deviation value reflects the matching error between the model prediction and the actual output. Before outputting the instantaneous deviation value to the dynamic parameter identification unit, data verification is performed to realize transmission integrity and provide high-reliability input for parameter adjustment.

[0038] Specifically, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The dynamic parameter identification unit receives the instantaneous deviation value, generates a weight factor greater than 0.8 to output an adjusted gain parameter in an initial stage, and generates a weight factor less than 0.3 to output an adjusted time constant parameter in a subsequent stage.

[0039] The dynamic parameter identification unit receives the instantaneous deviation value, generates a weight factor greater than a preset threshold to output an adjusted gain parameter in an initial stage, and generates a weight factor less than the preset threshold to output an adjusted time constant parameter in a subsequent stage. The weight factor is in a linear proportional relationship with the square of the deviation value, and the phased output strategy solves the contradiction between speed and accuracy in the parameter adjustment process: the high-weight stage prioritizes stabilizing system response, the low-weight stage refines steady-state accuracy, and the output parameter stream maintains the time sequence correlation of the gain parameter and the time constant parameter.

[0040] Specifically, the internal model control system based on real-time adaptive online identification of an industrial system according to the present application further comprises: The internal model control model updating unit receives the adjusted gain parameter and the time constant parameter, reconstructs a transfer function model by using a variable parameter transfer function function block, and updates the parameters in response to load fluctuations in the main steam pressure control of the circulating fluidized bed unit.

[0041] The internal model control model updating unit receives the adjusted gain parameter and the time constant parameter, reconstructs a transfer function model by using a variable parameter transfer function function block. In the main steam pressure control of the circulating fluidized bed unit, the model updates the parameters in response to load fluctuations: the gain parameter compensates for valve characteristic drift, and the time constant parameter optimizes the dynamic response speed. The reconstructed model is output to the controller parameter optimization unit to support real-time adaptation of the control algorithm to working condition changes.

[0042] Specifically, the internal model control system based on real-time adaptive online identification of industrial systems also includes: The controller parameter optimization unit receives the updated transfer function model and target set value in the main steam pressure control of the circulating fluidized bed unit, and outputs a continuously adjusted actuator opening degree instruction value to the control execution unit in the shaft seal temperature reduction control, so that the actual controlled system output value responds to the instruction change.

[0043] The controller parameter optimization unit receives the updated transfer function model and target set value in the main steam pressure control of the circulating fluidized bed unit, and outputs a continuously adjusted actuator opening degree instruction value to the control execution unit in the shaft seal temperature reduction control. Based on the transfer function model, dynamic weighting calculation is performed, the instruction value eliminates temperature control overshoot and oscillation, and the actual controlled system output value smoothly tracks the target set value.

[0044] Specifically, the internal model control system based on real-time adaptive online identification of industrial systems also includes: The control execution unit receives the actuator opening degree instruction value output in the shaft seal temperature reduction control, and drives the regulating valve in the supercritical boiler main control system, so that the error of the actual output value of the steam pressure tracking the target set value tends to zero.

[0045] The control execution unit receives the actuator opening degree instruction value output in the shaft seal temperature reduction control, and drives the regulating valve in the supercritical boiler main control system. The drive signal adjusts the valve opening through the output function block of the distributed control system, so that the error of the actual output value of the steam pressure tracking the target set value tends to zero. The execution result is fed back to the industrial process to form a closed-loop control flow from instruction to response.

[0046] Specifically, the internal model control system based on real-time adaptive online identification of industrial systems also includes: A bidirectional decoupling unit is arranged between the controller parameter optimization unit and the control execution unit, and the bidirectional decoupling unit performs: Receiving the actuator opening degree instruction value output by the first controller parameter optimization unit and the actuator opening degree instruction value output by the second controller parameter optimization unit; Generating the decoupled first actuator opening degree instruction value and the second actuator opening degree instruction value through cross parameter compensation; Outputting the decoupled first actuator opening degree instruction value to the first control execution unit and the second actuator opening degree instruction value to the second control execution unit.

[0047] A bidirectional decoupling unit is disposed between the controller parameter optimization unit and the control execution unit. The bidirectional decoupling unit receives the actuator opening degree instruction value output by the first controller parameter optimization unit and the actuator opening degree instruction value output by the second controller parameter optimization unit, and generates decoupled first and second actuator opening degree instruction values through cross parameter compensation. The decoupled instruction values are output to the first and second control execution units respectively, eliminating the bidirectional coupling interference of the multivariable system.

[0048] Specifically, the bidirectional decoupling unit of the internal model control system based on real-time adaptive online identification of industrial systems is configured to: identify the first gain parameter and the first time constant parameter from the actual controlled system output value fed back to the first control execution unit; identify the second gain parameter and the second time constant parameter from the actual controlled system output value fed back to the second control execution unit; input the first gain parameter and the first time constant parameter into the first internal model control model update unit; input the second gain parameter and the second time constant parameter into the second internal model control model update unit.

[0049] The bidirectional decoupling unit simultaneously performs identification of the first gain parameter and the first time constant parameter from the actual controlled system output value fed back to the first control execution unit, and identification of the second gain parameter and the second time constant parameter from the actual controlled system output value fed back to the second control execution unit. The first gain parameter and the first time constant parameter are input into the first internal model control model update unit, and the second gain parameter and the second time constant parameter are input into the second internal model control model update unit. The parameter transfer process maintains a grouping mapping relationship, supporting independent model updating of the decoupled system.

[0050] Specifically, the internal model control system based on real-time adaptive online identification of industrial systems further comprises: The bidirectional decoupling unit in the main steam pressure and primary air pressure dual control system of the circulating fluidized bed unit of the power plant updates the first gain parameter, the first time constant parameter, the second gain parameter, and the second time constant parameter in real time in response to load fluctuations.

[0051] The bidirectional decoupling unit in the main steam pressure and primary air pressure dual control system of the circulating fluidized bed unit of the power plant updates the first gain parameter, the first time constant parameter, the second gain parameter, and the second time constant parameter in real time in response to load fluctuations. The update operation is associated with the load fluctuation intensity: strong fluctuation scenarios increase the parameter adjustment frequency, and weak fluctuation scenarios reduce the adjustment amplitude, so that the decoupled system dynamically adapts to changes in bidirectional coupling.

[0052] The application solves the model mismatch problem caused by the time-varying characteristics of industrial processes through a synergistic mechanism. A real-time data acquisition unit obtains actuator opening command values, actual controlled system output values, and model controlled system output values from a distributed control system to construct an initial data source for feedback control. A deviation calculation unit calculates the deviation value between the actual controlled system output value and the model controlled system output value to quantify the instantaneous error between the model prediction and the actual state.

[0053] A dynamic parameter identification unit receives the deviation value and generates a weight factor, which is in a linear proportional relationship with the square of the deviation value. When the weight factor is greater than a preset threshold, only the adjusted gain parameter is output, and when the weight factor is less than the preset threshold, the adjusted time constant parameter is output. The phased adjustment mechanism prioritizes stabilizing system response and then refining steady-state accuracy. For example, in the main steam pressure control of a circulating fluidized bed unit, the gain parameter compensates for the characteristic drift caused by valve aging, and the time constant parameter optimizes the dynamic response under load fluctuations.

[0054] An internal model control model updating unit receives the updated gain parameter and time constant parameter to reconstruct the transfer function model. The reconstruction process uses a variable parameter transfer function function block to enable the model to adapt to changes in industrial processes in real time. In the temperature reducing water control scenario of a power plant boiler, the reconstructed model eliminates the transfer function drift caused by high temperature environments, and outputs a dynamic model to a controller parameter optimization unit.

[0055] The controller parameter optimization unit fuses the target set value, the actuator opening command value, and the updated transfer function model to generate an optimized actuator opening command value. In the shaft seal temperature reduction control, continuous adjustment instructions are output based on the dynamic model to eliminate temperature control overshoot. The control execution unit drives the regulating valve to execute the instructions, so that the actual controlled system output value responds to the instruction changes. The error between the actual output value of the supercritical boiler main control system steam pressure and the target set value tends to zero, forming a closed-loop control flow.

[0056] For a bidirectional coupling system, a bidirectional decoupling unit connects two control execution units. The unit receives the instruction values output by the first controller parameter optimization unit and the second controller parameter optimization unit, and generates decoupling instruction values through cross-parameter compensation. In the dual control of the main steam pressure and the primary air pressure of a circulating fluidized bed in a power plant, the bidirectional decoupling unit simultaneously identifies the gain parameters and time constant parameters of the two groups, updates the parameters in real time in response to load fluctuations, and eliminates the interference of bidirectional coupling.

[0057] The application forms a complete technical chain for dealing with time-varying characteristics through phased adjustment of parameters driven by weight factors, real-time reconstruction of variable parameter models, and decoupling control to eliminate coupling effects.

[0058] Please refer to Figures 2 to 9The embodiment of the present application is as follows: online identification implementation process based on DCS: using the basic operation function block of the distributed control system platform to realize online identification of the transfer function of the controlled object. Real-time collection of the actuator opening degree instruction value (OY), the actual controlled system output value (PV) and the model controlled system output value (IPV) is performed, and the deviation value (E) of PV and IPV is calculated. A linear relationship between the weight factor (L) and the square of the deviation value is established: L controls the parameter adjustment speed: when L is larger, the adjustment process is accelerated, and when L is smaller, the adjustment precision is slowed down; L controls the decoupling sequence of the parameters: when the initial L is larger, the gain parameter (K) is adjusted first, and when L becomes smaller, the time constant parameter (T) is adjusted; The identification process continuously optimizes the K and T parameters, so that the deviation value E gradually approaches zero. When the OY instruction changes, the IPV value gradually approaches the PV value, and the K and T parameters tend to be stable and meet the actual controlled system characteristics.

[0059] IMC internal model control fixed parameter model integrated control implementation process: in the existing single-input single-output internal model control architecture, the K and T parameters obtained by online identification are integrated, and a variable parameter transfer function function block is used to calculate the actuator opening degree instruction value (OY) in real time, and the model transfer function is updated to a real-time changing function; For time-varying objects such as power plant boiler desuperheating water control, boiler steam pressure control, shaft seal temperature reduction control, etc., the target set value (SP) is input into the integrated system. The online identification result (K, T) updates the internal model control model parameters and the controller parameters in real time, and when SP changes, OY is continuously adjusted to make IPV approach PV value, and the control quality is significantly better than that of the existing fixed parameter internal model control.

[0060] MIMO two-way decoupling control implementation process: for a multi-element system with two-way coupling (such as power plant circulating fluidized bed main steam pressure and primary air pressure double control), a multi-input multi-output two-way decoupling internal model control architecture is designed: A control structure of double target set values (SP1 / SP2), double controlled output values (PV1 / PV2) and double actuator opening degrees (OY1 / OY2) is constructed; Two-way disturbance correction algorithm is introduced through OY1 and OY2: OY1 affects PV1 and PV2, and OY2 affects PV2 and PV1; The inverse model of the two-way disturbance identification transfer function is introduced to correct the PV deviation in real time and realize decoupling; When the model is mismatched, SP1 / SP2 changes to trigger OY1 / OY2 continuous adjustment, and PV1 / PV2 stable tracking target value.

[0061] Online identification and decoupling are implemented simultaneously: online identification and decoupling control of multivariable system are executed synchronously, OY1 / OY2 simultaneously performs online identification on PV1 / PV2 process, and obtains two sets of parameters (K1, T1) and (K2, T2), which are integrated into two sets of internal model control models and controllers in real time. In the double control of circulating fluidized bed main steam pressure and wind pressure, the system identification and decoupling control are executed synchronously: when the load fluctuates, the parameters K1 / T1 / K2 / T2 are updated in real time, the bidirectional coupling interference is eliminated, and the control effect is significantly better than that of the scheme without online identification.

Claims

1. An internal model control system based on real-time adaptive online identification of industrial systems, characterized in that, include; The real-time data acquisition unit acquires the actuator opening command value, the actual output value of the controlled system, and the output value of the model controlled system. The deviation calculation unit receives the actual output value of the controlled system and the model output value of the controlled system from the real-time data acquisition unit, and outputs the deviation value between the actual output value of the controlled system and the model output value of the controlled system. The dynamic parameter identification unit receives the deviation value output by the deviation calculation unit, generates a weighting factor that is linearly proportional to the square of the deviation value, outputs the adjusted gain parameter when the weighting factor is greater than a preset threshold, and outputs the adjusted time constant parameter when the weighting factor is less than the preset threshold. The internal model control model update unit receives the gain parameter and time constant parameter output by the dynamic parameter identification unit, and outputs the reconstructed transfer function model. The controller parameter optimization unit receives the actuator opening command value, target set value, and transfer function model output by the internal model control model update unit from the real-time data acquisition unit, and outputs the optimized actuator opening command value. The control execution unit receives the optimized actuator opening command value output by the controller parameter optimization unit, and outputs a drive signal to the actuator so that the actual controlled system output value responds to the change of the optimized actuator opening command value.

2. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 1, characterized in that, Also includes: The real-time data acquisition unit, deviation calculation unit, and dynamic parameter identification unit are all completed through the function blocks of the distributed control system, and the parameter identification period is ≤200ms. The steam temperature data is obtained from the desuperheating water system of the power plant boiler.

3. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 2, characterized in that, Also includes: When the system is deployed on an external control device, the parameter identification cycle is extended to ≥1 second; The deviation calculation unit receives the steam temperature data output by the real-time data acquisition unit, uses a subtraction function block to calculate the instantaneous deviation between the temperature setpoint and the actual measured value in the denitrification control system, and outputs the instantaneous deviation value to the dynamic parameter identification unit.

4. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 3, characterized in that, Also includes: The dynamic parameter identification unit receives the instantaneous deviation value, generates a weight factor greater than 0.8 in the initial stage, outputs the adjusted gain parameter, and generates a weight factor less than 0.3 in the subsequent stage, outputs the adjusted time constant parameter.

5. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 4, characterized in that, Also includes: The internal model control model update unit receives the adjusted gain parameters and time constant parameters, reconstructs the transfer function model using a variable parameter transfer function block, and updates the parameters in response to load fluctuations in the main steam pressure control of the circulating fluidized bed unit.

6. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 5, characterized in that, Also includes: In the main air pressure control of the circulating fluidized bed unit, the controller parameter optimization unit receives the updated transfer function model and target setpoint, and outputs continuously adjustable actuator opening command values. In the shaft seal de-temperature control, it outputs continuously adjustable actuator opening command values ​​to the control execution unit, so that the actual controlled system output value responds to the command changes.

7. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 6, characterized in that, Also includes: The control execution unit receives the actuator opening command value output from the shaft seal desuperheating control, drives the regulating valve in the supercritical boiler main control system, and the error of the actual steam pressure output value tracking the target set value approaches zero.

8. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 7, characterized in that, Also includes: A bidirectional decoupling unit is deployed between the controller parameter optimization unit and the control execution unit, and the bidirectional decoupling unit performs the following: Receive the actuator opening command value output by the first controller parameter optimization unit and the actuator opening command value output by the second controller parameter optimization unit; The decoupled first actuator opening command value and second actuator opening command value are generated by cross parameter compensation; The decoupled first actuator opening command value is output to the first control execution unit, and the second actuator opening command value is output to the second control execution unit.

9. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 8, characterized in that, The bidirectional decoupling unit is configured as follows: Identify the first gain parameter and the first time constant parameter from the actual output value of the controlled system fed back by the first control execution unit; Identify the second gain parameter and the second time constant parameter from the actual output value of the controlled system fed back by the second control execution unit; The first gain parameter and the first time constant parameter are input into the first internal model control model update unit; The second gain parameter and the second time constant parameter are input into the second internal model control model update unit.

10. The internal model control system based on real-time adaptive online identification of industrial systems according to claim 9, characterized in that, Also includes: The bidirectional decoupling unit, in the dual control system of main steam pressure and primary air pressure of the circulating fluidized bed unit in the power plant, updates the first gain parameter, the first time constant parameter, the second gain parameter, and the second time constant parameter in real time in response to load fluctuations.

Images