Supervised multi-element embedding control method and system

Abstract

The invention relates to the field of thermal management control response systems, in particular to a supervised multi-element embedding control method and system.The method comprises the steps that self-inspection is conducted before a control response system is started; entering an internal adjusting process; the control response system transmits part of parameters to the discrete learning control unit and establishes a discrete parameter control model; establishing a prediction model equation according to the diagnosed control parameters, wherein the prediction model equation supervises, corrects and learns source data of the control response system; and establishing a mathematical model of a system stability module, and evaluating the state level of the system. According to the invention, the PID controller is not directly optimized, the control parameters and the system state parameters are normalized and discretized, and then a discrete parameter learning control module is designed, so that the stability of the system can be considered while the target parameters are controlled. Not only can the PID controller be directly applied, but also self-defined logic can be inserted into the system to participate in control.

Description

Technical Field

[0001] The present invention relates to the field of thermal management control response systems, and in particular to a supervisory multivariate embedded control method and system. Background Art

[0002] At present, due to the flexibility and reliability of PID controllers, about 95% of controllers in the field of industrial automation choose to use PID controllers. In the field of thermal management, PID controllers are also widely used. Not only that, many workers have also optimized PID controllers to varying degrees in order to adapt to the control response system in their respective fields. At present, embedded control and multivariate control are involved in various fields, but most of them are based on a specific system, such as lithium battery heating control. This control method has limitations for the control logic of new energy thermal management systems.

[0003] For example, in the existing technical patents, the patent application with publication number CN115774389A discloses a fuzzy PID control method and the patent application with publication number CN110520803A discloses a smart PID control method. It is undeniable that the above two control methods provide good ideas for optimizing PID controllers, but the optimization of these two control methods is based on PID, which limits the intervention of other logic units to a certain extent, thus affecting the efficiency and accuracy of the system; at the same time, for the coordinated development of thermal management systems, they ignore the actual state of the system and cannot reflect and adapt to the actual state of the control response system, thus affecting the accuracy of the system. Moreover, these two control methods only consider a single control parameter, without coordinating the action frequency of each control parameter, and cannot achieve the goal of efficient operation of the thermal management loop and new energy thermal management. In addition, these two control methods have flaws in ensuring the safety of the thermal management system loop.

[0004] Therefore, the control method of the prior art has the problems of poor accuracy, low working efficiency of the thermal management loop, and low safety. Summary of the invention

[0005] The purpose of the present invention is to provide a supervised multivariate embedded control method and system, which can solve the technical problems of poor accuracy, low working efficiency of thermal management loop and low safety of the control method in the prior art.

[0006] To achieve the above object, the present invention designs a supervised multivariate embedded control method, comprising the following steps:

[0007] The control response system performs a self-test before starting;

[0008] After the control response system completes startup, it enters the internal adjustment process;

[0009] Multiple logic units are embedded in the control response system, including a discrete learning control unit. The control response system transmits some parameters to the discrete learning control unit and establishes a discrete parameter control model.

[0010] When the discrete learning control unit completes learning and obtains the diagnosed control parameters, a prediction model equation is established according to the diagnosed control parameters, and the prediction model equation supervises, corrects and learns the source data of the control response system;

[0011] Establish a mathematical model of the system stability module, diagnose the system status, evaluate the system status level, and adapt different control parameters according to different system status levels;

[0012] Complete multivariate adaptive control of the control response system.

[0013] As a preferred solution, the self-check process before the control response system is started includes the following steps:

[0014] The control response system enters internal diagnosis after receiving the external start command, and then sends the system diagnosis flag to the supervision and feedback adaptation control unit for diagnosis;

[0015] When the monitoring and feedback adaptation control unit receives the diagnostic mark, it processes the feedback parameters and normalized parameters of the controlled object from the control response system, and uses an adder to accumulate and output the system diagnostic mark;

[0016] Determine whether the system diagnostic flag is within the safety threshold range for system startup. If the diagnosis passes, the startup of the control response system is completed; if the diagnosis fails, return to the previous step to continue diagnosis.

[0017] Further, judging whether the system diagnostic flag is within the safety threshold range of the system startup specifically includes the following steps:

[0018] Assume that there are two calibration parameters A and B, and a safety threshold Up;

[0019] SysDtcFlag = A*KSFb+B*Uni;

[0020] Among them, SysDtcFlag is the system diagnosis flag; KSFb is the system state coefficient, which is determined by the internal system state; Uni is calculated by normalizing the external input parameters;

[0021] In static state, Uni is very small, so it can be considered that SysDtcFlag = A*KSFb, that is, under static conditions, whether the control response system is started depends on KSFb; when the system state is unstable, KSFb is used to preliminarily reduce the frequency of system oscillation caused by blind startup of the controller;

[0022] When SysDtcFlag-Up≤0, the system is allowed to start; when SysDtcFlag-Up>0, it returns to the previous step to continue diagnosis.

[0023] As a preferred solution, the internal adjustment process of the control response system includes the following steps:

[0024] The internal adjustment system outputs dynamic adjustment target parameters, open-loop self-adjustment target parameters, initialization adjustment target parameters, and adjustment parameters respectively;

[0025] Dynamically adjust the target parameters, open-loop self-adjust the target parameters, and initialize the adjustment target parameters for weighted processing, and then perform weighted processing with the adjustment parameters;

[0026] The results are output to the control response system.

[0027] As a preferred solution, the establishment of the discrete parameter control model includes the following steps:

[0028] Define a valid control parameter NY, then there exists:

[0029] NY = CY + ((CY - Y) % 2β) * β;

[0030] Among them, CY is the discrete control parameter; Y is the unit control parameter; β is the response accuracy of the controlled object, which is a known quantity;

[0031] There is a functional relationship between the unit control parameter Y and the unit target parameter X: Y(X)=X;

[0032] Among them, X(n+1) is the future control parameter; X(n) is the current control parameter; X(n-1) is the past control parameter; y(n+1) is the future response parameter; y(n) is the current response parameter; y(n-1) is the past response parameter;

[0033] Then the following relationship exists in the control response system:

[0034] When |y(n+1)-y(n)|≥1, then KY=y(n+1)-1;

[0035] Among them, KY is the control parameter after diagnosis;

[0036] That is, when the change trend of the unit control parameter is greater than 1, it is effective control. At this time, the discrete learning control unit stores the effective data records here into a one-dimensional array; when y(n+1) infinitely approaches the value of y(n) rounded plus one, the discrete learning control unit will continue to refresh and record the data of X(n); when the value of y(n+1)-1 is greater than the value of y(n+1), the discrete learning control unit will repeat the above learning operation at the new address until all data is stored.

[0037] As a preferred solution, the method of establishing a prediction model according to the system state parameters and supervising, correcting and learning the source data of the control response system specifically includes the following steps:

[0038] When the system unit control parameter Y is updated, the unit control parameter Y is transmitted to the discrete learning control unit for determination through the diagnosed control target parameter TX. The determination rule is:

[0039] First, determine whether the unit control parameter Y falls within the region of the discrete learning control unit array; if so, determine the minimum interval of the region; if not, return to the initial state;

[0040] Secondly, the next state of the system is inferred based on the value of the unit control parameter Y, and it is determined whether the control parameter KY after diagnosis is reasonable; if it is reasonable, the weighting coefficient is mainly based on the unit control parameter Y; if it is unreasonable, the weighting coefficient is mainly based on the discrete control parameter CY.

[0041] Furthermore, the following mathematical model is used to determine whether the control parameter KY after diagnosis is reasonable:

[0042] If the known unit target parameter X and unit control parameter Y fall into a certain area TX[X1,X2] and KY[Y1,Y2] of the discrete learning control unit, the control target parameter TX and the control parameter KY after diagnosis are solved according to the known initial state (a,b) of the input parameters and the current output unit control parameter Y;

[0043] Take the maximum value X3 of a and X1, and the minimum value Y3 of b and Y2 to form a new interval [X3,X2], [Y1,Y3];

[0044] Using the linear approximation concept, when the new interval difference is very small, the irregular trajectory formed by the control parameters is approximately a straight line, that is, y = kx + C;

[0045] Obtain C=Y3-K*X2 ；

[0046] The system prediction model equation is obtained:

[0047]

[0048] When the value of a certain control parameter x is known, the predicted value of the prediction model equation y can be obtained; if the calculated predicted value y is better than the value in the learning array of the discrete learning control unit, that is, the control parameter KY after diagnosis is reasonable, then the weighting coefficient is mainly based on the unit control parameter Y; otherwise, it is mainly based on the discrete control parameter CY;

[0049] When the diagnosed control parameter KY successfully acts on the controlled object, the reasonableness of the diagnosed control parameter KY is determined according to the feedback state of the controlled object and the system stability parameter; if it is too large or too small, the value in the learning array of the discrete learning control unit is updated autonomously according to the scaling logic for next use;

[0050] Repeat the above operation, and with the continuous correction and learning of the control parameter KY after diagnosis, the conversion of the continuous control parameter to the discrete parameter can be completed.

[0051] As a preferred solution, the mathematical model of the system stability module is established to diagnose the system status and evaluate the system status level, which specifically includes the following steps:

[0052] Obtain various input parameters of the system stability module;

[0053] Converted into normalized parameter output;

[0054] Perform weighted processing on the output normalized parameters;

[0055] The normalized parameter is compared with the threshold value, and the system diagnosis level is output; if the system diagnosis level is 1, the control unit integrates quickly and outputs the control parameter; if the system diagnosis level is 2, the control unit integrates slowly and outputs the control parameter; if the system diagnosis level is 3, the control unit down-frequency discretizes and outputs the control parameter;

[0056] inputting the output control parameters into the response control unit;

[0057] The response control unit outputs the feedback parameter and continues the cycle as the input parameter.

[0058] Furthermore, the determination of the system diagnosis level specifically includes the following steps:

[0059] Assume that there are parameters Y1, Y2, Y3, Y4, and KS so that the unit control parameter Y and the control parameter KY after diagnosis have the following mathematical logic:

[0060] Y=KY=Y4*KS+Y4;

[0061] Y4=SysDtcLevel*(Y1+Y2+Y3);

[0062] Among them, KS is the adjustment frequency; Y1 is the dynamic adjustment coefficient; Y2 is the self-adjustment coefficient; Y3 is the initialization adjustment coefficient; Y4 is the target parameter set; SysDtcLevel is the system diagnosis level;

[0063] If and only if the system is stationary, Y1 = Y2 = Y3 = 0, otherwise there exists a unique parameter Y4 such that Y4 = Y1 or Y4 = Y2 or Y4 = Y3;

[0064] When SysDtcLevel=1, the unit control parameter Y after weighted processing is output to the control response system with a trend of fast integration and slight oscillation; when SysDtcLevel=2, the unit control parameter Y after weighted processing is output to the control response system with a trend of slow integration; when SysDtcLevel=3, the unit control parameter Y after weighted processing is output to the control response system with a trend of frequency reduction, discretization, safety and low speed.

[0065] The present invention also designs a supervised multivariate embedded control response system, including a self-checking module, an internal adjustment module, a discrete parameter control module, a prediction model equation module, a system state diagnosis module, and an adaptive control module;

[0066] The self-check module is used to control the self-check before the response system is started;

[0067] The internal adjustment module is used to control the response system to enter the internal adjustment process after the system is started up;

[0068] The discrete parameter control module is established to control the mutual embedding of multiple logic units in the response system, including a discrete learning control unit, to control the response system to pass some parameters to the discrete learning control unit and establish a discrete parameter control model;

[0069] The prediction model equation establishment module is used to obtain the diagnosed control parameters after the discrete learning control unit completes learning, and establish the prediction model equation according to the diagnosed control parameters, and the prediction model equation supervises, corrects and learns the source data of the control response system;

[0070] The system status diagnosis module is used to establish a mathematical model of the system stability module, diagnose the system status, evaluate the system status level, and adapt different control parameters according to different system status levels;

[0071] The adaptive control module is used to complete the multivariate adaptive control of the control response system.

[0072] Beneficial effects of the present invention:

[0073] Compared with the existing PID controller, the present invention does not directly optimize the PID controller, but is committed to normalizing and discretizing the control parameters and system state parameters, and then designs a discrete parameter learning control module, so that it can complete the control target parameters while also taking into account the stability of the system. The dynamic adjustment system and the static adjustment system are not completely coupled, not only can the PID controller be directly applied, but also custom logic can be inserted into the system to participate in the control. Since the present invention mainly learns and corrects the control parameters, it also means that the learning controller of the present invention will not be bound to the inherent control logic, thereby enriching the diversification and scalability of the control unit.

[0074] By combining the internal controller, system status diagnosis, discrete learning controller and embedded dynamic control unit, and adopting the intelligent open-loop and closed-loop control switching method, the system oscillation caused by the change of control parameters of the thermal management system can be reduced, and the efficient operation of the thermal management components can be ensured. At the same time, the present invention can also take into account the stability of the system while ensuring that the control response system can work normally. The defects of the existing control response system can be greatly improved.

[0075] By adding system status diagnosis, the controller can correct unreasonable control parameters in a timely manner, reducing the additional load on the thermal management system caused by mindless output, thereby improving the efficiency and accuracy of the system.

[0076] By adding a discrete learning control module and establishing a prediction model, the trend brought by the control parameters is predicted, and then the existing control parameters are corrected, so that the control parameters of the controlled object are more reasonable and accurate, thereby improving the accuracy of the system. In addition, the storage unit adopts a discrete distribution concept, which makes the control parameters of the learning control response system controllable and reduces the consumption of computing resources, so that the thermal management loop works efficiently.

[0077] Through the embedded dynamic control unit solution, the control parameters are simplistic, and the action frequency of each control parameter is coordinated, which improves the efficiency of the coordinated work of the control response system. In addition, the unit embedded control method provides more expansion interfaces while ensuring that the core control logic remains unchanged, enriches the control logic, and makes the control unit more complete and safer. Multiple control units restrict each other. When the dynamic adjustment system is out of adjustment, the control system will automatically start the protection measures. The hierarchical open-loop and supervisory control methods can maintain the stability of the system and ensure its safety during brainless high-power output. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a flow chart of the present invention.

[0079] Figure 2This is a control logic diagram of the system stability module of the present invention.

[0080] Figure 3 This is a control logic diagram of a controller in the prior art.

[0081] Figure 4 It is a control logic diagram of the controller of the present invention.

[0082] Figure 5 This is a control logic diagram of the controller self-check unit of the present invention.

[0083] Figure 6 for Figure 5 Control logic diagram of the system state coefficients.

[0084] Figure 7 This is the control logic diagram of the internal regulation system of the present invention.

[0085] Figure 8 It is a relationship diagram between the controlled parameters and the target parameters of the present invention.

[0086] Description of reference numerals:

[0087] TargetValue: target value; Sub: subtractor; Err: error value; KP: proportional value; KI: integral value; KD: differential value; ADD: adder; Y: unit control parameter; RealS: current state of the controlled object; Fb: feedback value; ErrLast: last difference; KI Last: last integral value; p: proportional coefficient; i: integral coefficient; d: differential coefficient;

[0088] Prci: low pressure; Prco: high pressure; Trci: low pressure temperature; Trco: high pressure temperature; TW: water temperature; IAMB: scene temperature of system operation; Res: calibration parameter; Uni: normalization parameter; SysDtcFlag: system diagnosis flag; SysCheckFlag: system detection flag; SysDtcLevel: system diagnosis level; SysFb: feedback parameter of controlled object; Y: unit control parameter; KY: control parameter after diagnosis; X: unit target parameter; TX: control target parameter after diagnosis; β is the response accuracy of the controlled object; CY is the discrete control parameter; KSFb: system state coefficient. DETAILED DESCRIPTION

[0089] In order to make the technical problems solved by the present invention, the technical solutions adopted and the technical effects achieved clearer, the technical solutions of the present invention are further described below in conjunction with the accompanying drawings and through specific implementation methods. It is understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, not all.

[0090] The present invention is mainly to solve the limitations of the thermal management control response system in the field of new energy, and is committed to solving and coordinating the coordination work between thermal management components. It mainly provides safe and reliable brain control for thermal management components such as compressors, electronic expansion valves, water pumps, and electronic fans. The control method mentioned in the present invention is no longer limited to a certain system, so that various thermal management control parameters can better coordinate and allocate resources, thereby working more efficiently. The present invention adds a system status diagnosis and discrete learning control unit, which can take into account the stability of the system while ensuring that the control response system can work normally, and can greatly improve the defects of the existing control response system.

[0091] The present invention relates to a supervised multivariate embedded control method and system. Compared with the existing PID controller, the present invention does not directly optimize the PID controller, but is committed to normalizing and discretizing the control parameters and system state parameters, and then designs a discrete parameter learning control module, so that it can complete the control target parameters while also taking into account the stability of the system. The dynamic adjustment system and the static adjustment system are not fully coupled, not only can the PID controller be directly applied, but also custom logic can be inserted into the system to participate in the control. Since the present invention mainly performs learning and correction on the control parameters, it also means that the learning controller of the present invention will not be bound to the inherent control logic, thereby enriching the diversification and scalability of the control unit.

[0092] The present invention patent introduces a supervised multi-element embedded control response system, which includes an adjustable self-starting control unit, a multi-level open-loop control unit, an embedded closed-loop control unit, a supervision and feedback adaptation control unit, a system stability module, a control response system, and a discrete learning control unit. The present invention uses multiple logic units to embed each other to achieve adaptive control of the target. Under the logic of normal control, the control parameters can be predicted, learned, and supervised and corrected. Among them, the adjustable self-starting control unit initially adjusts the starting speed of the component according to the properties of the controlled object, and then normalizes it according to the associated feedback parameters to select the level of the control parameter. The embedded closed-loop control unit in the invention is no longer unique. This module does not require the use of PID control, fuzzy control, robust control, optimal control, etc. with high security and practicality. Therefore, it can be replaced by table lookup method, linear fitting and other methods according to the adaptability of the system. At the same time, compared with the existing PID controller and other control response systems, they focus on the connection between the controlled object and the control parameters, but reduce the weight of system stability. This causes the existing control response system to always output according to one logic regardless of the system, which is not conducive to maintaining system stability and cannot make the controller work efficiently.

[0093] as follows Figure 3 As shown in FIG. 1 , the control logic diagram of the PID controller, which is widely used in the existing control field, can output a control parameter Y through mathematical operation to achieve control based on the difference between the control target and the actual value. Figure 3 The parameters are defined as follows:

[0094] TargetValue: target value, Sub: subtractor, Err: error value, KP: proportional value, KI: integral value, KD: differential value, ADD: adder, Y: unit control parameter, RealS: current state of the controlled object, Fb: feedback value, ErrLast: last difference, KI_Last: last integral value, p: proportional coefficient, i: integral coefficient, d: differential coefficient.

[0095] Through the following logic Figure 3 It can be known that the mathematical logic of this PID controller is:

[0096] Y=KP+KI+KD, where KP=Err*p, KI=∑Err*i, KD=(Err-ErrLast)*d;

[0097] Err = TargetValue - Fb;

[0098] Therefore, through the analysis of mathematical logic, it can be known that the logic of this PID controller is to use the difference to perform trend logic operations on the controlled object. It uses the ideas of proportion, integration and differentiation to well control the controlled object within the required target range.

[0099] However, existing PID controllers only focus on the relationship between the controlled object and the control target, ignoring other system parameters and the stability of the system. In a thermal management system, if you only focus on one control parameter, there will often be a scenario where the control parameter is always substandard but the PID controller keeps working, resulting in the inability of the controlled object to make correct adjustments when the system oscillates. At the same time, the existing thermal management control response systems are all coordinated by a set of core controllers, and the rest of the control logic is auxiliary. Therefore, when the control logic changes, the PID controller sometimes cannot perfectly adapt to the oscillation caused by the change, so it also needs to be modified to varying degrees as the control logic changes.

[0100] Therefore, in order to adapt to the coordinated control required in the field of thermal management and improve the single control of the current mainstream control, the present invention aims to break through the barriers of multiple control flows and adapt to the coordinated work requirements of multiple components in the field of thermal management, and at the same time aims to solve the problem of resource allocation of the existing control response system, and the unitized control method and system provided by the present invention can switch between different control logics according to needs. In addition, the present invention monitors the state of the system in real time by adding a system stability module and a supervisory unit block, and also diagnoses the system state by establishing a mathematical model, evaluates the state level of the system, and adapts different control parameters through different state levels, taking into account the stability of the system under the premise of coordinating multiple control targets, thereby ensuring the safety, reliability and efficiency of the entire thermal management control response system, while reducing the oscillation frequency of the system.

[0101] like Figure 4 As shown, the present invention designs a control logic diagram of a supervised multi-element embedded control response system, which includes a system control unit composed of an adjustable self-starting control unit, a multi-level open-loop control unit, an embedded closed-loop control unit, a supervision and feedback adaptation control unit, a system stability module, a control response system, and a discrete learning control unit. Figure 4 It is a distribution diagram of the entire thermal management control unit, which are coupled to each other and can selectively call individual sub-control units according to the weighted coefficients of the diagnostic parameters.

[0102] The adjustable self-start control unit receives the system status judgment from the supervision and feedback adaptation control unit. If the system status is good, the self-start control parameter is output to the control response system, otherwise the current status is maintained and the next instruction of the supervision and feedback adaptation control unit is awaited;

[0103] After completing the self-starting operation, the system will enter the multi-level open-loop control unit and wait for the system level judgment from the supervision and feedback adaptation control unit. Different initial control parameters will be output to the control response system according to different levels. After completing the open-loop control or receiving the instruction to enter the closed-loop control from the supervision and feedback adaptation control unit, the multi-level open-loop control unit will maintain the current state until the threshold of system stability is reached and then enter the embedded closed-loop control unit.

[0104] The embedded closed-loop control unit includes but is not limited to PID, robust control, custom control and other dynamic adjustment control response systems involved in the system. At the same time, this unit also receives the system status judgment from the supervision and feedback adaptation control unit in real time. The original control parameters will be disturbed only when the system is abnormal or it is judged that interference is needed. Otherwise, it will be output to the control response system normally.

[0105] The supervision and feedback adaptation control unit is at the core of the system. It not only coordinates the workflow of each unit, but also needs to judge the system stability module in real time. It then normalizes the various feedback parameters and outputs different system states according to different unit blocks. At the same time, this unit will record multiple sets of control data according to the changing state of the control target, and then perform differentiated intervention after each parameter calculation between the multi-level open-loop control unit and the embedded closed-loop control unit. After receiving the updated feedback parameters, it will compare them with the predicted parameters, and then take the optimal values ​​and update them to the previously stored data group. In addition, this unit will also judge the feasibility of the control parameters. If the output parameters of a certain state are significantly different from the recorded parameters due to changes in the dynamic control response system, intervention will be made, thereby assuming the risks brought about by the changes to a certain extent.

[0106] The complete implementation steps of the controller in the present invention are shown in the flowchart Figure 1 , Figure 1 A supervised multivariate embedded control method of the present invention comprises the following steps:

[0107] Step 1: The controller will undergo a self-test before starting. The detailed self-test unit is as follows Figure 5 As shown,

[0108] like Figure 5As shown, KSFb: system state coefficient. After receiving the external start command, the controller will enter the internal diagnosis, and then send the SysCheckFlag system diagnosis flag to the supervision and feedback adaptation control unit for diagnosis. When the supervision and feedback adaptation control unit receives the diagnosis flag, it will process the parameters SysFb and Uni normalized parameters from the control response system. Here, the adder is used to accumulate and output SysDtcFlag, and then diagnose whether SysDtcFlag is within the safety threshold of the startup. If the diagnosis passes, the startup is completed.

[0109] Therefore, combined with the above description, there will be the following mathematical logic:

[0110] Assume that there are two calibration parameters A and B, and a safety threshold Up;

[0111] SysDtcFlag = A*KSFb+B*Uni;

[0112] Then when SysDtcFlag-Up≤0, it is allowed to start. KSFB is determined by the internal system state. Uni is calculated by normalizing the external input parameters. In static state, Uni is very small, so it can be considered that SysDtcFlag=A*KSFb;

[0113] So the logic of KSFb is further analyzed, such as Figure 6 shown.

[0114] like Figure 6 As shown, the meanings of various parameters are as follows: KV: voltage coefficient; KI: current coefficient; Per: permission mark; Kr: calibration coefficient; V: voltage; I: current; WorkPermit: work permit; Rev: working system calibration parameters.

[0115] Then, there exist parameters KV, KA, Per, Kr, so that KSFb has the following relationship:

[0116] KSFb=V*KV+I*KI+Rev*Kr;

[0117] Among them, when Per is not permitted, KSFb is 0, otherwise it is calculated from various parameters.

[0118] Therefore, under static conditions, whether the controller starts depends on KSFb, that is, compared with the existing controller, a self-starting unit is pre-placed. When the system state is unstable, the frequency of system oscillation caused by the blind start of the controller can be preliminarily reduced. Under dynamic conditions, that is, when the controller starts working, three control units are built into the system automatic control of the present invention, namely, an adjustable self-starting control unit, a multi-level open-loop control unit, and an embeddable closed-loop control unit, which can work together or work alone.

[0119] Step 2: After the system is started, it will enter the internal adjustment system. The specific implementation method is as follows: Figure 7 After the controller is started, three logic blocks will be involved in the adjustment in the first stage, including the following steps:

[0120] The internal regulation system outputs DynaTarget dynamic regulation target parameter, SlefTarget open-loop self-regulation target parameter, InitCtr initialization regulation target parameter, and SlefFr regulation parameter respectively;

[0121] DynaTarget dynamic adjustment target parameter, SlefTarget open-loop self-adjustment target parameter, InitCtr initialization adjustment target parameter are weighted and then weighted with SlefFr adjustment parameter;

[0122] The results are output to the control response system.

[0123] Step 3: Multiple logic units in the control response system are embedded in each other, including the discrete learning control unit. When the system control unit is working, some parameters will be passed to the discrete learning control unit and a discrete parameter control model will be established;

[0124] like Figure 4 , which is a control logic diagram of the controller of the present invention;

[0125] Figure 4 The meanings of the parameters are as follows:

[0126] Prci: low pressure; Prco: high pressure; Trci: low pressure temperature; Trco: high pressure temperature; TW: water temperature; IAMB: scene temperature of system operation; Res: calibration parameter; Uni: normalization parameter; SysDtcFlag: system diagnosis flag; CellWorkSta: unit working status; SysCheckFlag: system detection flag; SysDtcLevel: system diagnosis level; SysFb: controlled object feedback parameter; Y: unit control parameter; KY: control parameter after diagnosis; TX: control target parameter after diagnosis; X: unit target parameter.

[0127] Define an effective control parameter NY, control the response system response accuracy β, then there exists

[0128] NY = CY + ((CY - Y) % 2β) * β;

[0129] That is, with the addition of the learning control unit, the control parameter NY input to the control response system is obtained by mathematical operation of the unit control parameter Y and the discrete control parameter CY. Among them, the learning control unit is a one-dimensional array with a length of 20, CY is the discrete control unit obtained by discrete operation of the one-dimensional table composed of the input parameters KY and TX, and the parameter Y is Figure 7 The parameters calculated by the internal control unit, β is the accuracy of the controlled object. Since β is a property of the controlled object, it can be considered a known quantity. Therefore, there is a certain mathematical relationship between CY and KY and TX.

[0130] Depend on Figure 4 It can be seen that KY is the control parameter after diagnosis, and TX is the control target parameter after diagnosis. Assume that the controlled parameter and the target parameter exist Figure 8 Relation Y(X)=X;

[0131] exist Figure 8 In , the meanings of various coordinate parameters are as follows:

[0132] X(n+1): future control parameter; X(n): current control parameter; X(n-1): past control parameter; y(n+1): future response parameter; y(n): current response parameter; y(n-1): past response parameter;

[0133] Then, in the system, there is the following relationship:

[0134] When |y(n+1)-y(n)|≥1, then KY=y(n+1)-1;

[0135] That is, when the change trend of the control target is greater than 1, it is considered to be effective control. At this time, the discrete learning control unit will store the effective data records here into a one-dimensional array; when y(n+1) infinitely approaches the value of y(n) rounded plus one, the discrete learning control unit will continue to refresh and record the data of X(n); when the value of y(n+1)-1 is greater than the value of y(n+1) when storing KY, the discrete learning control unit will repeat the above learning operation at the new address until 20 sets of data are stored.

[0136] Step 4: After the discrete learning control unit completes one learning, it obtains the system state parameters, and establishes a prediction model according to the system state parameters, and the prediction model supervises, corrects and learns the source data of the control response system;

[0137] When the system unit control parameter Y is updated, the unit control parameter Y is transmitted to the discrete learning control unit for determination through the diagnosed control target parameter TX. The determination rule is:

[0138] First, determine whether the unit control parameter Y falls within the region of the discrete learning control unit array; if so, determine the minimum interval of the region; if not, return to the initial state;

[0139] Secondly, the next state of the system is inferred based on the value of the unit control parameter Y, and whether the control parameter KY after diagnosis is reasonable; if it is reasonable, the weighting coefficient is mainly based on the unit control parameter Y; if it is unreasonable, the weighting coefficient is mainly based on the discrete control parameter CY;

[0140] The following is a simple mathematical model description of the above process:

[0141] If the known unit target parameter X and unit control parameter Y fall into a certain area TX[X1,X2] and KY[Y1,Y2] of the discrete learning control unit, the control target parameter TX and the control parameter KY after diagnosis are solved according to the known initial state (a,b) of the input parameters and the current output unit control parameter Y;

[0142] 1. Take the maximum value X3 of a and X1, and the minimum value Y3 of b and Y2 to form a new interval [X3, X2], [Y1, Y3]

[0143] 2. Using the linear approximation concept, when the difference between the new intervals is very small, the irregular trajectory formed by the control parameters is approximately a straight line, that is, y = kx + C;

[0144] 3. Then, we can get C=Y3-K*X2 ；

[0145] 4. Then, the system prediction model equation is obtained:

[0146]

[0147] 5. When the value of a certain control parameter x is known, the predicted value of the prediction model equation y can be obtained; if the calculated predicted value y is better than the value in the learning array of the discrete learning control unit, that is, the control parameter KY after diagnosis is reasonable, then the weighting coefficient is mainly based on the unit control parameter Y; otherwise, it is mainly based on the discrete control parameter CY;

[0148] 6. When the diagnosed control parameter KY successfully acts on the controlled object, determine whether the diagnosed control parameter KY is reasonable based on the feedback state of the controlled object and the system stability parameter; if it is too large or too small, the value in the learning array of the discrete learning control unit is updated autonomously according to the scaling logic for next use.

[0149] 7. Repeat the above steps 1-6. With the continuous correction and learning of the control parameter KY after diagnosis, the conversion of the continuous control parameter to the discrete parameter can be completed. That is, as the data of the discrete learning control unit becomes more and more accurate, the internal dynamic adjustment controller such as the PID controller can be released by adjusting the weighting coefficient, and the one-dimensional learning data can be used instead, thereby releasing some resources and reducing the burden on the controller.

[0150] Since the present invention mainly learns and modifies the control parameters, it also means that the learning controller of the present invention will not be bound to the inherent control logic, thereby enriching the diversification and scalability of the control unit. As the control parameters stabilize, the weight of the discrete learning control unit controller in the present invention can be gradually increased. Therefore, whether this control response system can maintain the stability of the system is also the focus of the present invention.

[0151] Step 5: Establish a mathematical model of the system stability module, diagnose the system status, evaluate the system status level, and adapt different control parameters according to different system status levels;

[0152] Step 5.1: Figure 2 It is the control logic diagram of the system stability module, which specifically includes the following steps:

[0153] Obtain various input parameters of the system stability module;

[0154] Converted into normalized parameter output;

[0155] Perform weighted processing on the output normalized parameters;

[0156] The normalized parameter is compared with the threshold value, and the system diagnosis level is output; if the system diagnosis level is 1, the control unit integrates quickly and outputs the control parameter; if the system diagnosis level is 2, the control unit integrates slowly and outputs the control parameter; if the system diagnosis level is 3, the control unit down-frequency discretizes and outputs the control parameter;

[0157] inputting the output control parameters into the response control unit;

[0158] The response control unit outputs the feedback parameter and continues the cycle as the input parameter.

[0159] Depend on Figure 4 It can be seen that the input of the system stability module consists of Prci low pressure; Prco high pressure; Trci low pressure temperature; Trco high pressure temperature; TW water temperature; IAMB system working scene temperature; Res calibration parameters; SysFb controlled object feedback parameters; the output is Uni normalized parameters, so the following mathematical model can be established about Uni normalized parameters:

[0160] There are calibrated weighting coefficients Pi, Po, Ti, To, Tw, IA, KR, Ks such that:

[0161] Uni=Pi*Prci+Po*Prco+Ti*Trci+To*Trco+(∑Tw*TW)+IAMB*IA+Res*KR+Ks*SysFb;

[0162] The meanings of the calibration parameters are as follows:

[0163] Pi: low pressure weighting coefficient; Po: high pressure weighting coefficient; Ti: low pressure temperature weighting coefficient; To: high pressure temperature weighting coefficient; Tw: water temperature weighting coefficient; IA: system working environment temperature weighting coefficient; KR: calibration coefficient; Ks: controlled object feedback parameter weighting coefficient.

[0164] For each input parameter, there is a limit value in the actual thermal management control response system, so a general formula for normalization can be given: Taking the low pressure Prci as an example, assuming that the limits of the control parameters are [MIN, MAX], then the low pressure Prci has a normalized value;

[0165] Right now

[0166] Similarly, the remaining parameters Prco, Trci, Trco, TW, IAMB, Res, and SysFb can also be normalized using this logic.

[0167] Step 5.2: The determination of the system diagnostic level specifically includes the following steps:

[0168] Assume that there are parameters Y1, Y2, Y3, Y4, KS so that the following mathematical logic exists:

[0169] Y=KY=Y4*KS+Y4;

[0170] Y4=SysDtcLevel*(Y1+Y2+Y3);

[0171] Among them, DynaTarget: dynamic adjustment target parameter; SlefTarget: open-loop self-adjustment target parameter; InitCtr: initialization adjustment target parameter; SlefFr: adjustment parameter; Y1: dynamic adjustment coefficient; Y2: self-adjustment coefficient; Y3: initialization adjustment coefficient; Y4: target parameter set; KS: adjustment frequency;

[0172] And there exists the following logic: Y1, Y2, Y3 if and only if the system is stationary Y1=Y2=Y3=0; otherwise there exists a unique parameter Y4 such that Y4=Y1 or Y4=Y2 or Y4=Y3.

[0173] When SysDtcLevel=1, the unit control parameter Y after weighted processing is output to the control response system with a trend of fast integration and slight oscillation; when SysDtcLevel=2, the unit control parameter Y after weighted processing is output to the control response system with a trend of slow integration; when SysDtcLevel=3, the unit control parameter Y after weighted processing is output to the control response system with a trend of frequency reduction, discretization, safety and low speed.

[0174] Therefore, it can be concluded that as the diagnostic level variable SysDtcLevel changes, the controller core also changes. However, due to the static and dynamic hybrid module, the core can be fixed and dynamically adjusted, or the core can be switched for control as SysDtcLevel changes. This can be used to prevent the problem of control system maladjustment caused by unclear system boundaries in the early stage.

[0175] Therefore, compared with the existing PID controller, the present invention adds SysDtcLevel system diagnostic level judgment in the control. The addition of this diagnosis effectively prevents the problem of controller imbalance caused by unclear initial system boundaries. At the same time, since the three built-in adjustment systems are mutually restricted, when the dynamic adjustment system such as the PID controller is out of adjustment, the system will automatically start the protection measures. In addition, it can be concluded from the above description that the dynamic adjustment system of the present invention is not completely coupled with the static adjustment system, so when applying the system designed by the present invention, not only can the PID regulator be directly added, but also customized logic such as customized table coefficients can be inserted to participate in the control.

[0176] In order to better achieve accurate system response, the calibration parameters can be corrected by empirical table lookup compensation. When all parameters are determined, the final output normalized parameter Uni is judged and compared with the threshold to evaluate the system level. Finally, each control unit executes different control unit logic according to different levels.

[0177] Step 6: Complete the multivariate adaptive control of the control response system.

[0178] Therefore, in the present invention, the PID controller is not optimized directly, but the control parameters and system state parameters are normalized and discretized, and then a discrete parameter learning control module is designed, so that it can complete the control target parameters while taking into account the stability of the system.

[0179] It should be understood that the specific order or hierarchy of steps in the process disclosed by the present invention is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of protection of the present disclosure. The attached method claims provide the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.

[0180] The present invention also provides a supervised multivariate embedded control response system, including a self-checking module, an internal adjustment module, a discrete parameter control module, a prediction model equation module, a system state diagnosis module, and an adaptive control module;

[0181] The self-check module is used to control the self-check before the response system is started;

[0182] The internal adjustment module is used to control the response system to enter the internal adjustment process after the system is started up;

[0183] The discrete parameter control module is established to control the mutual embedding of multiple logic units in the response system, including a discrete learning control unit, to control the response system to pass some parameters to the discrete learning control unit and establish a discrete parameter control model;

[0184] The prediction model equation establishment module is used to obtain the diagnosed control parameters after the discrete learning control unit completes learning, and establish the prediction model equation according to the diagnosed control parameters, and the prediction model equation supervises, corrects and learns the source data of the control response system;

[0185] The system status diagnosis module is used to establish a mathematical model of the system stability module, diagnose the system status, evaluate the system status level, and adapt different control parameters according to different system status levels;

[0186] The adaptive control module is used to complete the multivariate adaptive control of the control response system.

[0187] like Figure 4 As shown in the figure, the control logic diagram of the system includes an adjustable self-starting control unit, a multi-level open-loop control unit, an embedded closed-loop control unit, and a supervision and feedback adaptation control unit. The system is characterized in that a self-starting unit is pre-installed in the system. When the system state is unstable, the frequency of system oscillation can be slowed down. At the same time, oscillation factor judgment is added to the control. When the stability of the system exceeds the control target, the system will automatically start the protection measures, which is conducive to the safe startup of a certain control target unit.

[0188] In the system open-loop control, there are three levels, namely low-speed safety start, smooth safety start, and fast slight oscillation start. The judgment of the three levels is obtained based on the normalization of various feedback parameters of the system. In addition, unlike the existing open-loop technology, in this invention, the control target does not need to complete all open-loop controls before switching to closed-loop control. As long as the system status allows and the switching threshold is reached, it can be switched. This operation not only improves the fault tolerance of the system, but also reduces the time required to judge whether the components enter the logic control in different environments to a certain extent. It ensures the efficient operation of electrical components in different environments.

[0189] Closed-loop system designers do not need to be constrained by the compatibility of different controls. In this invention, the coupling of closed-loop control is greatly weakened, so that the closed-loop system can not only work normally, but also provide reference data for other units. At the same time, designers can also change other closed-loop control methods in the middle without major modifications. This ensures the fault tolerance and adjustability of the development project process.

[0190] In addition, the present invention also has a monitoring unit, which is used to monitor various parameters and perform mathematical operations, and then provide reference data for other units. At the same time, it is also the hub linking open-loop control and closed-loop control, and when the parameters of the control unit are abnormal, it will assist in correction, ensuring the efficiency and stability of the control logic.

[0191] The present invention also provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned supervisory multi-element embedded control method are implemented.

[0192] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0193] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A supervised multivariate embedded control method, characterized in that: The following steps are included: The control response system performs a self-test before starting; After the control response system completes startup, it enters the internal adjustment process; Multiple logic units are embedded in the control response system, including a discrete learning control unit. The control response system transmits some parameters to the discrete learning control unit and establishes a discrete parameter control model. When the discrete learning control unit completes learning and obtains the diagnosed control parameters, a prediction model equation is established according to the diagnosed control parameters, and the prediction model equation supervises, corrects and learns the source data of the control response system; Establish a mathematical model of the system stability module, diagnose the system status, evaluate the system status level, and adapt different control parameters according to different system status levels; Complete multivariate adaptive control of the control response system.

2. A supervised multivariate embedded control method according to claim 1, characterized in that: The self-check process before the control response system is started includes the following steps: The control response system enters internal diagnosis after receiving the external start command, and then sends the system diagnosis flag to the supervision and feedback adaptation control unit for diagnosis; When the monitoring and feedback adaptation control unit receives the diagnostic mark, it processes the feedback parameters and normalized parameters of the controlled object from the control response system, and uses an adder to accumulate and output the system diagnostic mark; Determine whether the system diagnostic flag is within the safety threshold range for system startup. If the diagnosis passes, the startup of the control response system is completed; if the diagnosis fails, return to the previous step to continue diagnosis.

3. A supervised multivariate embedded control method according to claim 2, characterized in that: Determining whether the system diagnostic flag is within the safe threshold range for system startup specifically includes the following steps: Assume that there are two calibration parameters A and B, and a safety threshold Up; SysDtcFlag = A*KSFb+B*Uni; Among them, SysDtcFlag is the system diagnosis flag; KSFb is the system state coefficient, which is determined by the internal system state; Uni is calculated by normalizing the external input parameters; In static state, Uni is very small, so it can be considered that SysDtcFlag = A*KSFb, that is, under static conditions, whether the control response system is started depends on KSFb; when the system state is unstable, KSFb is used to preliminarily reduce the frequency of system oscillation caused by blind startup of the controller; When SysDtcFlag-Up≤0, the system is allowed to start; when SysDtcFlag-Up>0, it returns to the previous step to continue diagnosis.

4. A supervised multivariate embedded control method according to claim 3, characterized in that: The control response system internal adjustment process includes the following steps: The internal adjustment system outputs dynamic adjustment target parameters, open-loop self-adjustment target parameters, initialization adjustment target parameters, and adjustment parameters respectively; Dynamically adjust the target parameters, open-loop self-adjust the target parameters, and initialize the adjustment target parameters for weighted processing, and then perform weighted processing with the adjustment parameters; The results are output to the control response system.

5. A supervised multivariate embedded control method according to claim 4, characterized in that: The discrete parameter control model is established, comprising the following steps: Define a valid control parameter NY, then there exists: NY = CY + ((CY - Y) % 2β) * β; Among them, CY is the discrete control parameter; Y is the unit control parameter; β is the response accuracy of the controlled object, which is a known quantity; There is a functional relationship between the unit control parameter Y and the unit target parameter X: Y(X)=X; Among them, X(n+1) is the future control parameter; X(n) is the current control parameter; X(n-1) is the past control parameter; y(n+1) is the future response parameter; y(n) is the current response parameter; y(n-1) is the past response parameter; Then the following relationship exists in the control response system: When |y(n+1)-y(n)|≥1, then KY=y(n+1)-1; Among them, KY is the control parameter after diagnosis; That is, when the change trend of the unit control parameter is greater than 1, it is effective control. At this time, the discrete learning control unit stores the effective data records here into a one-dimensional array; when y(n+1) infinitely approaches the value of y(n) rounded plus one, the discrete learning control unit will continue to refresh and record the data of X(n); when the value of y(n+1)-1 is greater than the value of y(n+1), the discrete learning control unit will repeat the above learning operation at the new address until all data is stored.

6. A supervised multivariate embedded control method according to claim 5, characterized in that: The method of establishing a prediction model according to the system state parameters and supervising, correcting and learning the source data of the control response system specifically includes the following steps: When the system unit control parameter Y is updated, the unit control parameter Y is transmitted to the discrete learning control unit for determination through the diagnosed control target parameter TX. The determination rule is: First, determine whether the unit control parameter Y falls within the region of the discrete learning control unit array; if so, determine the minimum interval of the region; if not, return to the initial state; Secondly, the next state of the system is inferred based on the value of the unit control parameter Y, and it is determined whether the control parameter KY after diagnosis is reasonable; if it is reasonable, the weighting coefficient is mainly based on the unit control parameter Y; if it is unreasonable, the weighting coefficient is mainly based on the discrete control parameter CY.

7. A supervised multivariate embedded control method according to claim 6, characterized in that: The following mathematical model is used to determine whether the control parameter KY after diagnosis is reasonable: If the known unit target parameter X and unit control parameter Y fall into a certain area TX[X1,X2] and KY[Y1,Y2] of the discrete learning control unit, the control target parameter TX and the control parameter KY after diagnosis are solved according to the known initial state (a,b) of the input parameters and the current output unit control parameter Y; Take the maximum value X3 of a and X1, and the minimum value Y3 of b and Y2 to form a new interval [X3,X2], [Y1,Y3]; Using the linear approximation concept, when the new interval difference is very small, the irregular trajectory formed by the control parameters is approximately a straight line, that is, y = kx + C; Obtain C = Y3-K*X2; The system prediction model equation is obtained: When the value of a certain control parameter x is known, the predicted value of the prediction model equation y can be obtained; if the calculated predicted value y is better than the value in the learning array of the discrete learning control unit, that is, the control parameter KY after diagnosis is reasonable, then the weighting coefficient is mainly based on the unit control parameter Y; otherwise, it is mainly based on the discrete control parameter CY; When the diagnosed control parameter KY successfully acts on the controlled object, the reasonableness of the diagnosed control parameter KY is determined according to the feedback state of the controlled object and the system stability parameter; if it is too large or too small, the value in the learning array of the discrete learning control unit is updated autonomously according to the scaling logic for next use; Repeat the above operation, and with the continuous correction and learning of the control parameter KY after diagnosis, the conversion of the continuous control parameter to the discrete parameter can be completed.

8. A supervised multivariate embedded control method according to claim 7, characterized in that: The method of establishing a mathematical model of the system stability module, diagnosing the system status, and evaluating the system status level specifically includes the following steps: Obtain various input parameters of the system stability module; Converted into normalized parameter output; Perform weighted processing on the output normalized parameters; The normalized parameter is compared with the threshold value, and the system diagnosis level is output; if the system diagnosis level is 1, the control unit integrates quickly and outputs the control parameter; if the system diagnosis level is 2, the control unit integrates slowly and outputs the control parameter; if the system diagnosis level is 3, the control unit down-frequency discretizes and outputs the control parameter; inputting the output control parameters into the response control unit; The response control unit outputs the feedback parameter and continues the cycle as the input parameter.

9. A supervised multivariate embedded control method according to claim 8, characterized in that: The determination of the system diagnostic level specifically includes the following steps: Assume that there are parameters Y1, Y2, Y3, Y4, and KS so that the unit control parameter Y and the control parameter KY after diagnosis have the following mathematical logic: Y=KY=Y4*KS+Y4; Y4=SysDtcLevel*(Y1+Y2+Y3); Among them, KS is the adjustment frequency; Y1 is the dynamic adjustment coefficient; Y2 is the self-adjustment coefficient; Y3 is the initialization adjustment coefficient; Y4 is the target parameter set; SysDtcLevel is the system diagnosis level; If and only if the system is stationary, Y1 = Y2 = Y3 = 0, otherwise there exists a unique parameter Y4 such that Y4 = Y1 or Y4 = Y2 or Y4 = Y3; When SysDtcLevel=1, the unit control parameter Y after weighted processing is output to the control response system with a trend of fast integration and slight oscillation; when SysDtcLevel=2, the unit control parameter Y after weighted processing is output to the control response system with a trend of slow integration; when SysDtcLevel=3, the unit control parameter Y after weighted processing is output to the control response system with a trend of frequency reduction, discretization, safety and low speed.

10. A supervised multivariate embedded control response system, characterized in that: It includes a self-check module, an internal adjustment module, a discrete parameter control module, a prediction model equation module, a system status diagnosis module, and an adaptive control module; The self-check module is used to control the self-check before the response system is started; The internal adjustment module is used to control the response system to enter the internal adjustment process after the system is started up; The discrete parameter control module is established to control the mutual embedding of multiple logic units in the response system, including a discrete learning control unit, to control the response system to pass some parameters to the discrete learning control unit and establish a discrete parameter control model; The prediction model equation establishment module is used to obtain the diagnosed control parameters after the discrete learning control unit completes learning, and establish the prediction model equation according to the diagnosed control parameters, and the prediction model equation supervises, corrects and learns the source data of the control response system; The system status diagnosis module is used to establish a mathematical model of the system stability module, diagnose the system status, evaluate the system status level, and adapt different control parameters according to different system status levels; The adaptive control module is used to complete the multivariate adaptive control of the control response system.

Images