CN106842934A - A kind of SPC and IMC methods of two input and output network decoupling and controlling system random delay - Google Patents

Abstract

SPC the and IMC methods of two input and output network decoupling and controlling system (TITO NDCS) random delay, belong to the MIMO NDCS technical fields of limited bandwidth resources.Affect one another and couple between a kind of two input/output signal, need the TITO NDCS by decoupling treatment, transmit produced network delay among the nodes due to network data, not only influence the stability of respective close loop control circuit, but also the stability of whole system will be influenceed, even result in the problem that TITO NDCS lose stabilization, propose with the network data transmission process between all real nodes in TITO NDCS, instead of the method for network delay compensation model therebetween, SPC and IMC are implemented respectively to two loops, the measurement to network delay between node can be exempted, estimate or recognize, reduce clock signal synchronization requirement, random delay is reduced to TITO NDCS stability influences, improve quality of system control.

Description

A kind of SPC and IMC methods of two input and output network decoupling and controlling system random delay

Technical field

A kind of SPC of two input and output network decoupling and controlling system random delay (Smith Predictor Control, SPC) and IMC (Internal Model Control, IMC) method, it is related to automatic control technology, the network communications technology and calculating The crossing domain of machine technology, more particularly to limited bandwidth resources MIMO Networked Control Systems technical field.

Background technology

In dcs, sensor and controller, between controller and actuator, by Real Time Communication Network The closed-loop feedback control system of composition, referred to as network control system (Networked control systems, NCS), NCS's Typical structure is as shown in Figure 1.

NCS is capable of achieving resource-sharing, remote operation and control, tool compared with the control system of traditional point-to-point structure There is a high diagnosis capability, I＆M is easy, many advantages, such as increased flexibility and the reliability of system.Long-range distant behaviour Work, telemedicine, remote teaching, wireless network robot, some Weapon Systems and emerging with fieldbus and industrial ether Control system based on net belongs to the category of NCS, additionally, NCS is in aerospace field, and complicated, dangerous industry Control field also has wide application, and it is studied has turned into a hot subject of international academic community.

In NCS, due to the presence of the phenomenons such as network delay, data packetloss and network congestion so that NCS faces many New challenge.Sensor as NCS, when passing through network exchange data between controller and actuator, when inevitably resulting in network Prolong, so as to the performance of system can be reduced or even cause system unstable.Because the information source in network is a lot, transmitting data stream warp Numerous computers and communication equipment and path is not exclusive；Or limitation and the influence of transmission mechanism due to the network bandwidth, network The reason such as congestion or disconnecting, causes the sequential entanglement of network packet and the loss of packet.Although time-delay system point Analysis and modeling obtained in recent years there may be in remarkable progress, but NCS various time delays of different nature (constant, bounded, with Machine, time-varying etc.) so that existing method typically can not be applied directly.Traditional control theory is being analyzed and is setting to system Timing, has often done many Utopian it is assumed that transmitting and adjusting such as the sampling of single rate, Synchronization Control, without time delay.But in NCS In, because control loop has network, above-mentioned hypothesis is typically invalid, therefore Traditional control theory will be reappraised Can be applied in NCS.

At present, the research on NCS both at home and abroad, primarily directed to single-input single-output (Single-input and Single-output, SISO) network control system, respectively known to network delay, it is unknown or random, network delay be less than one The individual sampling period transmits more than a sampling period, the transmission of list bag or many bags, when whetheing there is data-bag lost, it is entered Row mathematical modeling or stability analysis and controlling.But in actual industrial process, generally existing including at least two inputs Export the control system of (Two-input and two-output, TITO), the multiple-input and multiple-output (Multiple- for being constituted Input and multiple-output, MIMO) network control system research it is then relatively fewer, in particular for input with Between output signal, there is coupling needs by decoupling the multiple-input and multiple-output network decoupling and controlling system for processing (Networked decoupling control systems, NDCS) delay compensation with control achievement in research then it is relative more It is few.

The typical structure of MIMO-NDCS is as shown in Figure 2.

Compared with SISO-NCS, MIMO-NDCS has the characteristics that：

(1) affected one another between input signal and output signal and there is coupling

In the MIMO-NCS that there is coupling, a change for input signal will become multiple output signals Change, and each output signal is also not only influenceed by an input signal.Even if by meticulous between input and output signal Selection pairing, also exists and influences each other unavoidably between each control loop, thus it is respective output signal is independently tracked Input signal is had any problem.Decoupler in MIMO-NDCS, for releasing or reducing the coupling between MIMO signal Cooperation is used.

(2) internal structure is more more complex than SISO-NCS and MIMO-NCS

(3) controlled device there may be uncertain factor

In MIMO-NDCS, the parameter being related to is more, and the contact between each control loop is more, and parameter variations are to overall control The influence of effect processed can become very complicated.

(4) control unit failure

In MIMO-NDCS, including at least there is two or more close loop control circuits, including at least have two or More than two sensors and actuator.The failure of each element may influence the performance of whole control system, when serious Control system can be made unstable, or even caused a serious accident.

Due to the above-mentioned particularity of MIMO-NDCS so that be mostly based on SISO-NCS be designed with control method, The requirement of the control performance of MIMO-NDCS and control quality cannot have been met, prevent its from or be not directly applicable MIMO- In the design and analysis of NDCS, control and design to MIMO-NDCS bring certain difficulty.

For MIMO-NDCS, network delay compensation is essentially consisted in the difficult point of control：

(1) due to network delay and network topology structure, communication protocol, offered load, the network bandwidth and data package size It is relevant etc. factor, to more than several or even the dozens of sampling period random network time delay, to set up each control in MIMO-NDCS The Mathematical Modeling that the random network time delay in loop processed is accurately predicted, estimates or recognized, is nearly impossible at present.

(2) occur in MIMO-NDCS, when previous node is to network during latter node-node transmission network data Prolong, no matter using which kind of prediction or method of estimation in previous node, be impossible to know the net for producing thereafter in advance in advance The exact value of network time delay.Time delay causes systematic function to decline or even causes system unstable, while also to the analysis of control system Difficulty is brought with design.

(3) to meet in MIMO-NDCS, all node clock signal Complete Synchronizations in different distributions place are unrealistic 's.

(4) due in MIMO-NCS, being affected one another between input and output, and there is coupling, its MIMO-NDCS's Internal structure is more complicated than MIMO-NCS and SISO-NCS, it is understood that there may be uncertain factor it is more, to MIMO-NDCS implement Delay compensation is more much more difficult than MIMO-NCS and SISO-NCS with control.

The content of the invention

A kind of two input and output network decoupling and controlling system (TITO-NDCS) in the present invention relates to MIMO-NDCS is random The compensation of time delay and control, the typical structure of its TITO-NDCS are as shown in Figure 3.

For the close loop control circuit 1 in Fig. 3：

1) from input signal x₁S () arrives output signal y₁S the closed loop transfer function, between () is：

In formula：C₁S () is controller；G₁₁S () is controlled device；τ₁Represent the output signal u of controller C nodes₁(s), Through preceding the random network time delay that decoupling actuator DA1 nodes are experienced is transferred to network path；τ₂Represent sensor S1 sections The output signal y of point₁(s), through the random network time delay that feedback network tunnel is experienced to controller C nodes.

2) the uneoupled control signal u of actuator DA2 nodes is decoupled from close loop control circuit 2_p2(s), by cross decoupling Path transmission function P₁₂(s) and controlled device line passing transmission function G₁₂S () acts on close loop control circuit 1, believe from input Number u_p2S () arrives output signal y₁S the closed loop transfer function, between () is：

The denominator of above-mentioned closed loop transfer function, equation (1) to (2)In, when containing random network Prolong τ₁And τ₂Exponential termWithThe presence of time delay loses the performance quality of control system, the system of even resulting in is deteriorated Stability.

For the close loop control circuit 2 in Fig. 3：

1) from input signal x₂S () arrives output signal y₂S the closed loop transfer function, between () is：

In formula：C₂S () is controller, G₂₂S () is controlled device；τ₃Represent the controlled output signal u of controller C nodes₂ S (), the random network time delay that decoupling actuator DA2 nodes are experienced is transferred to through preceding to network path；τ₄Represent sensor The output signal y of S2 nodes₂(s), through the random network time delay that feedback network tunnel is experienced to controller C nodes.

2) the uneoupled control signal u of actuator DA1 nodes is decoupled from close loop control circuit 1_p1(s), by cross decoupling Path transmission function P₂₁(s) and controlled device line passing transmission function G₂₁S () acts on close loop control circuit 2, believe from input Number u_p1S () arrives output signal y₂S the closed loop transfer function, between () is：

The denominator of above-mentioned closed loop transfer function, equation (3) to (4)In, when containing random network Prolong τ₃And τ₄Exponential termWithThe presence of time delay loses the performance quality of control system, the system of even resulting in is deteriorated Stability.

Goal of the invention：

For the TITO-NDCS of Fig. 3, in the denominator of the closed loop transfer function, equation (1) to (2) of its close loop control circuit 1, Contain random network delay, τ₁And τ₂Exponential termWithAnd the closed loop transfer function, equation of close loop control circuit 2 (3) in the denominator of (4), random network delay, τ is contained₃And τ₄Exponential termWithThe presence of time delay can be reduced respectively From the control performance quality of close loop control circuit and the stability of respective close loop control circuit is influenceed, while will also decrease whole system The control performance quality of system simultaneously influences the stability of whole system, and whole system loss of stability will be caused when serious.

Therefore, for the close loop control circuit 1 in Fig. 3：The present invention proposes a kind of delay compensation method based on SPC；Pin To close loop control circuit 2：The present invention proposes a kind of delay compensation method based on IMC；When constituting two close loop control circuit networks The compensation prolonged with mix control, in exempting to each close loop control circuit, between node random network time delay measurement, estimate Or identification, and then reduce network delay τ₁And τ₂, and τ₃And τ₄To respective close loop control circuit and to whole control system control The influence of performance quality processed and the stability of a system；It is capable of achieving not including network delay in the characteristic equation of respective close loop control circuit Exponential term, and then influence of the network delay to whole system stability can be reduced, improve the dynamic property quality of system, realize Segmentation, real-time, online and dynamic predictive compensation and SPC and IMC to TITO-NDCS random network time delays.

Using method：

For the close loop control circuit 1 in Fig. 3：

The first step：In order to realize meeting during predictive compensation condition, the closed loop transform function of close loop control circuit 1 is no longer included Network delay exponential term, to realize to network random delay τ₁And τ₂Compensation with control, around controlled device G₁₁(s), to close Ring control loop 1 exports y₁(s) as input signal, by y₁S () passes through network transfer delay prediction modelAnd Prediction Control Device C_1m(s) and network transfer delay prediction modelOne positive feedback Prediction Control loop of construction, by y₁S () is by estimating Controller C_1mS () constructs a negative-feedback Prediction Control loop；The structure for implementing this step is as shown in Figure 4；

Second step：In for actual TITO-NDCS, it is difficult to obtain the problem of network delay exact value, to realize in fig. 4 Compensation and control to network delay, it is necessary to meet network delay prediction modelWithTo be equal to its true modelWithCondition, and meet predictor controller C_1mS () is equal to its controller C₁S the condition of () is (due to controller C₁S () is people To design and selecting, C is met naturally_1m(s)=C₁(s)).Therefore, from sensor S1 nodes to controller C nodes, and From controller C nodes to decoupling actuator DA1 nodes, using real network data transmission processWithInstead of it Between network delay predict-compensate modelWithObtain the compensation of network random delay and the control structure shown in Fig. 5；

3rd step：By controller C in Fig. 5₁S (), by the further abbreviation of transmission function equivalence transformation rule, obtains Fig. 6 institutes The network delay compensation of the implementation the inventive method shown and control structure；Realize system not comprising network delay therebetween from structure Predict-compensate model so that in exempting to close loop control circuit 1, network random delay τ between node₁And τ₂Measurement, estimate Or identification, it is capable of achieving to network random delay τ₁And τ₂Compensation and SPC；Implement the network random delay compensation of the inventive method It is as shown in Figure 6 with SPC structures.

For the close loop control circuit 2 in Fig. 3：

The first step：In controller C nodes, an internal mode controller C is built_2IMC(s) substitution controller C₂(s)；For reality When now meeting predictive compensation condition, the closed loop transform function of close loop control circuit 2 no longer includes network delay exponential term, to realize To network random delay τ₃And τ₄Compensation with control, around controlled device G₂₂S (), y is exported with close loop control circuit 2₂S () is made It is input signal, by y₂S () passes through network transfer delay prediction modelWith estimate internal mode controller C_2mIMC(s) and network Propagation delay time prediction modelOne positive feedback Prediction Control loop of construction；The structure for implementing this step is as shown in Figure 4；

Second step：In for actual TITO-NDCS, it is difficult to obtain the problem of network delay exact value, to realize in fig. 4 Compensation and control to network delay, it is necessary to meet network delay prediction modelWithTo be equal to its true modelWithCondition, and satisfaction estimate internal mode controller C_2mIMCS () is equal to its internal mode controller C_2IMCS the condition of () is (due to interior Mould controller C_2IMCS () is artificial design and selection, C is met naturally_2mIMC(s)=C_2IMC(s)).Therefore, from sensor S2 nodes To between controller C nodes, and from controller C nodes to decoupling actuator DA2 nodes, using real network data Transmitting procedureWithInstead of the predict-compensate model of network delay therebetweenWithThe network obtained shown in Fig. 5 is random Delay compensation and control structure；

3rd step：By internal mode controller C in Fig. 5_2IMCS (), by the further abbreviation of transmission function equivalence transformation rule, obtains The network delay compensation of the implementation the inventive method shown in Fig. 6 and control structure；Realize system not comprising net therebetween from structure The predict-compensate model of network time delay, so that in exempting to close loop control circuit 2, network random delay τ between node₃And τ₄Survey Amount, estimation are recognized, and are capable of achieving to network random delay τ₃And τ₄Compensation and IMC；When the network of implementation the inventive method is random Prolong compensation as shown in Figure 6 with IMC structures.

Herein it should be strongly noted that in the controller C nodes of Fig. 6, close loop control circuit 1 being occurred in that respectively and being returned The Setting signal x on road 2₁(s) and x₂(s), respectively with its feedback signal y₁(s) and y₂(s) implement first " subtracting " afterwards " plus ", or first " plus " operation rule that " subtracts " afterwards, i.e. y₁(s) and y₂S () signal is connected to controller C by positive feedback and negative-feedback simultaneously respectively In node：

(1) this is due to by the controller C in Fig. 5₁(s) and internal mode controller C_2IMC(s), respectively according to transmission function etc. The further abbreviation of valency transformation rule obtains the result shown in Fig. 6, and non-artificial setting；

(2) because the node of NCS is nearly all intelligent node, not only with communication and calculation function, but also with depositing Storage with control etc. function, same signal is carried out in node elder generation " subtracting " afterwards " plus ", or first " plus " " subtract " afterwards, this is in operation method Then go up do not have what be not inconsistent normally part；

Same signal is carried out in node (3) " plus " with " subtracting " computing its end value it is " zero ", this " zero " value, and The signal y in the node is not indicated that₁(s) or y₂S () does not just exist, or do not obtain y₁(s) or y₂(s) signal, or signal It is not stored for；Or do not exist because " cancelling out each other " causes " zero " signal value to reform into, or it is nonsensical；

(4) triggering of controller C nodes, just comes from signal y₁(s) or y₂The driving of (s), if controller C nodes It is not received by the signal y come from feedback network tunnel₁(s) or y₂(s), then in event-driven working method Controller C nodes will not be triggered.

For the close loop control circuit 1 in Fig. 6：

1) from input signal x₁S () arrives output signal y₁S the closed loop transfer function, between () is：

In formula：C₁S () is controller.

2) the signal u of decoupling actuator DA2 nodes in close loop control circuit 2 is come from_2p(s), by cross decoupling path Transmission function P₁₂S () acts on close loop control circuit 1；At the same time, signal u_2pS () is transmitted by controlled device line passing Function G₁₂S () acts on close loop control circuit 1；From input signal u_2pS () arrives output signal y₁Closed loop transfer function, between (s) For：

Using the inventive method, the closed loop transform function of close loop control circuit 1 is 1+C₁(s)G₁₁S ()=0, its closed loop is special Levy in equation no longer comprising the network random delay τ of the influence stability of a system₁And τ₂Exponential termWithSo as to reduce Influence of the network delay to the stability of a system, improves system dynamic control performance quality, realizes the dynamic to network random delay Compensation and SPC.

For the close loop control circuit 2 in Fig. 6：

1) from input signal x₂S () arrives output signal y₂S the closed loop transfer function, between () is：

In formula：C_2IMCS () is internal mode controller.

2) the signal u of decoupling actuator DA1 nodes in close loop control circuit 1 is come from_1p(s), by cross decoupling path Transmission function P₂₁S () acts on close loop control circuit 2；At the same time, signal u_1pS () is transmitted by controlled device line passing Function G₂₁S () acts on close loop control circuit 2；From input signal u_1pS () arrives output signal y₂Closed loop transfer function, between (s) For：

Using the inventive method, the denominator of transmission function equation (7) and (8) is 1, is no longer included in its closed loop transform function Influence the random network delay, τ of the stability of a system₃And τ₄Exponential termWithThe stability of system is only passed by controlled device Delivery function, cross decoupling path transmission function and internal mode controller are determined；So as to network delay can be reduced to system stabilization Property influence, improve system dynamic control performance quality, realize compensation to random network time delay and IMC.

In close loop control circuit 1, controller C₁The selection of (s)：

Controller C₁S () can be according to controlled device G₁₁The Mathematical Modeling of (s), and model parameter change, both may be selected Conventional control strategy, also may be selected Based Intelligent Control or complex control strategy；Close loop control circuit 1 uses SPC methods, from TITO- Realized in NDCS structures and specific controller C₁S the selection of the control strategy of () is unrelated.

In close loop control circuit 2, internal mode controller C_2IMCThe design of (s) and selection：

Design internal mode controller typically uses pole-zero cancellation method, i.e. two step design methods：The first step is that design one takes it It is the inversion model of plant model as feedforward controller C₂₂(s)；Second step is that certain order is added in feedforward controller Feedforward filter f₂S (), constitutes a complete internal mode controller C_2IMC(s)。

(1) feedforward controller C₂₂(s)

Error, the interference of system when first ignoring controlled device and plant model Incomplete matching and other are various about The factors such as beam condition, in selection close loop control circuit 2, controlled device prediction model is equal to its true model, i.e.,：G_22m(s)=G₂₂ (s)。

Now, controlled device prediction model can be divided into according to the poles and zeros assignment situation of controlled device：G_22m(s)= G_22m+(s)G_22m-(s), wherein：G_22m+S () is controlled device prediction model G_22mPure lag system and s RHPs are included in (s) The irreversible part of zero pole point；G_22m-S () is the reversible part of minimum phase in controlled device prediction model.

Under normal circumstances, the feedforward controller C of close loop control circuit 2₂₂S () can be chosen for：

(2) feedforward filter f₂(s)

The thing of feedforward controller can be influenceed due to the pure lag system in controlled device and positioned at the zero pole point of s RHPs Reason is realisation, thus the reversible part G of controlled device minimum phase has only been taken in the design process of feedforward controller_22m-(s), Have ignored G_22m+(s)；There is error due to possible Incomplete matching between controlled device and controlled device prediction model, system In there is likely to be interference signal, these factors are likely to make system to lose stabilization.Therefore, adding one in feedforward controller Determine the feedforward filter of order, for reducing influence of the factors above to the stability of a system, improve the robustness of system.

Generally the feedforward filter f of close loop control circuit 2₂S (), is chosen for fairly simple n₂Rank wave filterWherein：λ₂It is feedforward filter time constant；n₂It is the order of feedforward filter, and n₂=n_2a-n_2b；n_2a It is controlled device G₂₂The order of (s) denominator；n_2bIt is controlled device G₂₂The order of (s) molecule, usual n₂＞ 0.

(3) internal mode controller C_2IMC(s)

The internal mode controller C of close loop control circuit 2_2IMCS () can be chosen for:

Be can be seen that from equation (9)：The internal mode controller C of one degree of freedom_2IMCIn (s), the adjustable ginseng of only one of which Number λ₂, due to λ₂The change of parameter suffers from direct relation with the tracking performance of system and antijamming capability, therefore in filter of adjusting The customized parameter λ of ripple device₂When, the tracing property generally required in system is traded off between the two with antijamming capability.

The scope of application of the invention：

Suitable for known to plant model or a kind of two input and output network decoupling and controlling system (TITO- for being uncertain of NDCS) compensation of random network time delay and SPC and IMC；Its Research Thinking and research method, can equally be well applied to controlled device mould Type is known or the compensation of multiple-input and multiple-output network decoupling and controlling system (MIMO-NCS) random network time delay that is uncertain of with SPC and IMC.

It is a feature of the present invention that the method is comprised the following steps：

For close loop control circuit 1：

(1) is h when the sensor S1 nodes cycle₁Sampled signal trigger when, employing mode A is operated；

(2) is when controller C nodes are by feedback signal y₁When () triggers s, employing mode B is operated；

(3) is when decoupling actuator DA1 nodes are by signal e₁When () triggers s, employing mode C is operated；

For close loop control circuit 2：

(4) is h when the sensor S2 nodes cycle₂Sampled signal trigger when, employing mode D is operated；

(5) is when controller C nodes are by feedback signal y₂When () triggers s, employing mode E is operated；

(6) is when decoupling actuator DA2 nodes are by IMC signals u₂When () triggers s, employing mode F is operated；

The step of mode A, includes：

A1：Sensor S1 nodes work in time type of drive, and its trigger signal is cycle h₁Sampled signal；

A2：After sensor S1 nodes are triggered, to controlled device G₁₁The output signal y of (s)₁₁S () and controlled device are intersected Channel transfer function G₁₂The output signal y of (s)₁₂S () is sampled, and calculate the system output signal of close loop control circuit 1 y₁(s), and y₁(s)=y₁₁(s)+y₁₂(s)；

A3：By feedback signal y₁(s), by the feedback network path of close loop control circuit 1 to controller C node-node transmissions, Feedback signal y₁S () will experience network transfer delay τ₂Afterwards, controller C nodes are got to；

The step of mode B, includes：

B1：Controller C nodes work in event driven manner, by feedback signal y₁S () is triggered；

B2：In controller C nodes, by the system Setting signal x of close loop control circuit 1₁(s), with feedback signal y₁(s) After phase adduction subtracts each other, signal, i.e. e are obtained₁(s)=x₁(s)+y₁(s)-y₁(s)=x₁(s)；

B3：By signal e₁S feedforward network path that () passes through close loop control circuit 1Unit is saved to decoupling actuator DA1 Point transmission, e₁S () will experience network transfer delay τ₁Afterwards, get to decouple actuator DA1 nodes；

The step of mode C, includes：

C1：Decoupling actuator DA1 nodes work in event driven manner, by signal e₁S () is triggered；

C2：By signal e₁(s) and feedback signal y₁S () subtracts each other and obtains signal e₃(s), i.e. e₃(s)=e₁(s)-y₁(s)；It is right e₃S () implements control algolithm C₁S (), obtains control signal u₁(s)；

C3：By control signal u₁S () decouples the output signal u of actuator DA2 nodes with close loop control circuit 2 is come from_2p S () passes through cross decoupling path transmission function P₁₂The output signal y of (s)_p12S () subtracts each other and obtains signal u_1p(s), i.e. u_1p(s)= u₁(s)-y_p12(s)；

C4：By signal u_1pS () acts on controlled device G₁₁S () obtains its output valve y₁₁(s)；By signal u_1pS () acts on Controlled device cross aisle transmission function G₂₁S () obtains its output valve y₂₁(s)；So as to realize to controlled device G₁₁(s) and G₂₁ S the uneoupled control of () adds SPC, while realizing to network delay τ₁And τ₂Compensation with control；

The step of mode D, includes：

D1：Sensor S2 nodes work in time type of drive, and its trigger signal is cycle h₂Sampled signal；

D2：After sensor S2 nodes are triggered, to controlled device G₂₂The output signal y of (s)₂₂S () and controlled device are intersected Channel transfer function G₂₁The output signal y of (s)₂₁S () is sampled, and calculate the system output signal of close loop control circuit 2 y₂(s), and y₂(s)=y₂₂(s)+y₂₁(s)；

D3：By feedback signal y₂(s), by the feedback network path of close loop control circuit 2 to controller C node-node transmissions, Feedback signal y₂S () will experience network transfer delay τ₄Afterwards, controller C nodes are got to；

The step of mode E, includes：

E1：Controller C nodes work in event driven manner, by feedback signal y₂S () is triggered；

E2：In controller C nodes, by the system Setting signal x of close loop control circuit 2₂(s), with feedback signal y₂(s) phase After adduction subtracts each other, signal e is obtained₂(s), i.e. e₂(s)=x₂(s)+y₂(s)-y₂(s)=x₂(s)；

E3：To e₂S () implements Internal Model Control Algorithm C_2IMCS (), obtains IMC signals u₂(s)；

E4：By IMC signals u₂S feedforward network path that () passes through close loop control circuit 2Unit is to decoupling actuator DA2 node-node transmissions, u₂S () will experience network transfer delay τ₃Afterwards, get to decouple actuator DA2 nodes；

The step of mode F, includes：

F1：Decoupling actuator DA2 nodes work in event driven manner, by IMC signals u₂S () is triggered；

F2：By IMC signals u₂S () decouples the output signal u of actuator DA1 nodes with close loop control circuit 1 is come from_1p S () passes through cross decoupling path transmission function P₂₁The output signal y of (s)_p21S () subtracts each other and obtains signal u_2p(s), i.e. u_2p(s)= u₂(s)-y_p21(s)；

F3：By signal u_2pS () acts on controlled device G₂₂S () obtains its output valve y₂₂(s)；By signal u_2pS () acts on Controlled device cross aisle transmission function G₁₂S () obtains its output valve y₁₂(s)；So as to realize to controlled device G₂₂(s) and G₁₂ S the uneoupled control of () adds IMC, while realizing to network delay τ₃And τ₄Compensation with control.

The present invention has following features：

1st, due to from exempting in structure in TITO-NDCS, the measurement of network delay, observation, estimate or recognize, while also The synchronous requirement of node clock signal can be exempted, time delay can be avoided to estimate the inaccurate evaluated error for causing of model, it is to avoid pair when Prolong the waste of consuming node storage resources needed for identification, while can also avoid due to " the sky sampling " or " sampling more " that time delay is caused The compensation error brought.

2nd, it is unrelated with the selection of specific network communication protocol due to being realized from TITO-NDCS structures, thus be both applicable In the TITO-NDCS using wired network protocol, the TITO-NDCS of wireless network protocol is also applicable for use with；It is not only suitable for really Qualitative procotol, also suitable for the procotol of uncertainty；The TITO-NDCS of heterogeneous network composition is not only suitable for, while Also it is applied to the TITO-NDCS that heterogeneous network is constituted.

3rd, the control loop 1 in TITO-NDCS uses SPC, due to being realized from TITO-NDCS structures and specific controller C₁S the selection of the control strategy of () is unrelated, thus can be not only used for, using the TITO-NDCS of conventional control, also can be used to use intelligence Can control or the TITO-NDCS using complex control strategy.

4th, the control loop 2 in TITO-NDCS uses IMC, its internal mode controller C_2IMCThe adjustable parameter only one of which of (s) λ₂Parameter, the regulation of its parameter is simple with selection, and explicit physical meaning；Can not only be improved using IMC system stability, Tracking performance and interference free performance, but also the compensation to random network time delay and IMC can be realized.

5th, because the present invention uses compensation and control method that " software " changes TITO-NDCS structures, thus at it Any hardware device need not be further added by implementation process, the software resource carried using existing TITO-NDCS intelligent nodes, it is sufficient to Its compensation and control function are realized, hardware investment can be saved and be easy to be extended and applied.

Brief description of the drawings

Fig. 1：The typical structure of NCS

Fig. 1 is by sensor S nodes, controller C nodes, actuator A nodes, controlled device, feedforward network tunnel list UnitAnd feedback network tunnel unitConstituted.

In Fig. 1：X (s) represents system input signal；Y (s) represents system output signal；C (s) represents controller；U (s) tables Show control signal；τ_caThe feedforward network that control signal u (s) is experienced in expression from controller C nodes to actuator A node-node transmissions Tunnel time delay；τ_scThe feedback net that detection signal y (s) of sensor S nodes is experienced in expression to controller C node-node transmissions Network tunnel time delay；G (s) represents controlled device transmission function.

Fig. 2：The typical structure of MIMO-NDCS

Fig. 2 is by r sensor S node, controller C nodes, m decoupling actuator DA node, controlled device G, m forward direction Network path propagation delay timeUnit, and r feedback network tunnel time delayUnit Constituted.

In Fig. 2：y_jS () represents j-th output signal of system；u_iS () represents i-th control signal；Representing will control Signal u_i(s) from controller C nodes to i-th decoupling actuator DA node-node transmissions experienced feedforward network tunnel when Prolong；Represent j-th detection signal y of sensor S nodes_jS () leads to the feedback network that controller C node-node transmissions are experienced Road propagation delay time；G represents controlled device transmission function.

Fig. 3：The typical structure of TITO-NDCS

Fig. 3 is made up of close loop control circuit 1 and 2, and its system includes sensor S1 and S2 node, controller C nodes, solution Coupling actuator DA1 and A2 node, controlled device transmission function G₁₁(s) and G₂₂(s) and controlled device line passing transmission function G₂₁(s) and G₁₂(s), cross decoupling path transmission function P₂₁(s) and P₁₂(s), feedforward network tunnel unitWith And feedback network tunnel unitWithConstituted.

In Fig. 3：x₁(s) and x₂S () represents the input signal of system；y₁(s) and y₂S () represents the output signal of system；C₁ (s) and C₂S () represents the controller of control loop 1 and 2；u₁(s) and u₂S () represents control signal；τ₁And τ₃Represent and believe control Number u₁(s) and u₂S feedforward network tunnel that () is experienced from from controller C nodes to decoupling actuator DA1 and A2 node-node transmission Time delay；τ₂And τ₄Represent the detection signal y of sensor S1 and S2 node₁(s) and y₂S () is experienced to controller C node-node transmissions Feedback network tunnel time delay.

Fig. 4：A kind of TITO-NDCS delay compensations comprising prediction model and control structure

In Fig. 4：C_1mS () is the controller C of control loop 1₁The prediction model of (s)；C_2mIMCS () is the internal model control of control loop 2 Device C processed_2IMCThe prediction model of (s)；AndIt is network transfer delayAndEstimate Time Delay Model；With AndIt is network transfer delayAndEstimate Time Delay Model.

Fig. 5：Replace the delay compensation of prediction model and control structure with true model

Fig. 6：A kind of SPC and IMC methods of two input and output network decoupling and controlling system random delay

Specific embodiment

Exemplary embodiment of the invention will be described in detail by referring to accompanying drawing 6 below, make the ordinary skill people of this area Member becomes apparent from features described above of the invention and advantage.

Specific implementation step is as described below：

For close loop control circuit 1：

The first step：Sensor S1 nodes work in time type of drive, are h when the sensor S1 nodes cycle₁Sampling After signal triggering, will be to controlled device G₁₁The output signal y of (s)₁₁(s) and controlled device cross aisle transmission function G₁₂(s) Output signal y₁₂S () is sampled, and calculate the system output signal y of close loop control circuit 1₁(s), and y₁(s)=y₁₁(s) +y₁₂(s)；

Second step：Sensor S1 nodes are by feedback signal y₁(s), by the feedback network path of close loop control circuit 1 to Controller C node-node transmissions, feedback signal y₁S () will experience network transfer delay τ₂Afterwards, controller C nodes are got to；

3rd step：Controller C nodes work in event driven manner, by feedback signal y₁S () triggers after, with feedback letter Number y₁S () phase adduction subtracts each other after, signal e is obtained₁(s), i.e. e₁(s)=x₁(s)+y₁(s)-y₁(s)=x₁(s)；

4th step：By signal e₁S feedforward network path that () passes through close loop control circuit 1Unit is to decoupling actuator DA1 node-node transmissions, e₁S () will experience network transfer delay τ₁Afterwards, get to decouple actuator DA1 nodes；

5th step：Decoupling actuator DA1 nodes work in event driven manner, by signal e₁After (s) triggering, by signal e₁ (s) and feedback signal y₁S () subtracts each other and obtains signal e₃(s), i.e. e₃(s)=e₁(s)-y₁(s)；To e₃S () implements control algolithm C₁ S (), obtains control signal u₁(s)；By control signal u₁S () decouples actuator DA2 nodes with close loop control circuit 2 is come from Output signal u_2pS () passes through cross decoupling path transmission function P₁₂The output signal y of (s)_p12S () subtracts each other, obtain signal u_1p (s), i.e. u_1p(s)=u₁(s)-y_p12(s)；

6th step：By signal u_1pS () acts on controlled device G₁₁S () obtains its output valve y₁₁(s)；By signal u_1pS () is made For controlled device cross aisle transmission function G₂₁S () obtains its output valve y₂₁(s)；So as to realize to controlled device G₁₁(s) and G₂₁S the uneoupled control of () adds IMC, while realizing to network delay τ₁And τ₂Compensation with control；

7th step：Return to the first step；

For close loop control circuit 2：

The first step：Sensor S2 nodes work in time type of drive, are h when the sensor S2 nodes cycle₂Sampling After signal triggering, will be to controlled device G₂₂The output signal y of (s)₂₂(s) and controlled device cross aisle transmission function G₂₁(s) Output signal y₂₁S () is sampled, and calculate the system output signal y of close loop control circuit 2₂(s), and y₂(s)=y₂₂(s) +y₂₁(s)；

Second step：Sensor S2 nodes are by feedback signal y₂(s), by the feedback network path of close loop control circuit 2 to Controller C node-node transmissions, feedback signal y₂S () will experience network transfer delay τ₄Afterwards, controller C nodes are got to；

3rd step：Controller C nodes work in event driven manner, by feedback signal y₂S () triggers after, by closed loop control The system Setting signal x of loop processed 2₂(s), with feedback signal y₂S () phase adduction obtains signal e after subtracting each other₂(s), i.e. e₂(s)=x₂ (s)+y₂(s)-y₂(s)=x₂(s)；To e₂S () implements Internal Model Control Algorithm C_2IMCS (), obtains IMC signals u₂(s)；

4th step：By IMC signals u₂S feedforward network path that () passes through close loop control circuit 2Unit is performed to decoupling Device DA2 node-node transmissions, u₂S () will experience network transfer delay τ₃Afterwards, get to decouple actuator DA2 nodes；

5th step：Decoupling actuator DA2 nodes work in event driven manner, by IMC signals u₂S () triggers after, will IMC signals u₂S () decouples the output signal u of actuator DA1 nodes with close loop control circuit 1 is come from_1pS () passes through cross decoupling Path transmission function P₂₁The output signal y of (s)_p21S () subtracts each other, obtain signal u_2p(s), i.e. u_2p(s)=u₂(s)-y_p21(s)；

6th step：By signal u_2pS () acts on controlled device G₂₂S () obtains its output valve y₂₂(s)；By signal u_2pS () is made For controlled device cross aisle transmission function G₁₂S () obtains its output valve y₁₂(s)；So as to realize to controlled device G₂₂(s) and G₁₂S the uneoupled control of () adds IMC, while realizing to network delay τ₃And τ₄Compensation with control；

7th step：Return to the first step；

The foregoing is only presently preferred embodiments of the present invention and oneself, be not intended to limit the invention, it is all in essence of the invention Within god and principle, any modification, equivalent substitution and improvements made etc. should be included within the scope of the present invention.

The content not being described in detail in this specification belongs to prior art known to professional and technical personnel in the field.

Claims (4)

1. a kind of SPC and IMC methods of two input and output network decoupling and controlling system random delay, it is characterised in that the method bag Include following steps：

For close loop control circuit 1：

(1) is h when the sensor S1 nodes cycle₁Sampled signal trigger when, employing mode A is operated；

(2) is when controller C nodes are by feedback signal y₁When () triggers s, employing mode B is operated；

(3) is when decoupling actuator DA1 nodes are by signal e₁When () triggers s, employing mode C is operated；

For close loop control circuit 2：

(4) is h when the sensor S2 nodes cycle₂Sampled signal trigger when, employing mode D is operated；

(5) is when controller C nodes are by feedback signal y₂When () triggers s, employing mode E is operated；

(6) is when decoupling actuator DA2 nodes are by IMC signals u₂When () triggers s, employing mode F is operated；

The step of mode A, includes：

A1：Sensor S1 nodes work in time type of drive, and its trigger signal is cycle h₁Sampled signal；

A2：After sensor S1 nodes are triggered, to controlled device G₁₁The output signal y of (s)₁₁(s) and controlled device cross aisle Transmission function G₁₂The output signal y of (s)₁₂S () is sampled, and calculate the system output signal y of close loop control circuit 1₁ (s), and y₁(s)=y₁₁(s)+y₁₂(s)；

A3：By feedback signal y₁(s), by the feedback network path of close loop control circuit 1 to controller C node-node transmissions, feedback letter Number y₁S () will experience network transfer delay τ₂Afterwards, controller C nodes are got to；

The step of mode B, includes：

B1：Controller C nodes work in event driven manner, by feedback signal y₁S () is triggered；

B2：In controller C nodes, by the system Setting signal x of close loop control circuit 1₁(s), with feedback signal y₁S () is added And after subtracting each other, obtain signal, i.e. e₁(s)=x₁(s)+y₁(s)-y₁(s)=x₁(s)；

B3：By signal e₁S feedforward network path that () passes through close loop control circuit 1Unit is passed to decoupling actuator DA1 nodes It is defeated, e₁S () will experience network transfer delay τ₁Afterwards, get to decouple actuator DA1 nodes；

The step of mode C, includes：

C1：Decoupling actuator DA1 nodes work in event driven manner, by signal e₁S () is triggered；

C2：By signal e₁(s) and feedback signal y₁S () subtracts each other and obtains signal e₃(s), i.e. e₃(s)=e₁(s)-y₁(s)；To e₃(s) Implement control algolithm C₁S (), obtains control signal u₁(s)；

C3：By control signal u₁S () decouples the output signal u of actuator DA2 nodes with close loop control circuit 2 is come from_2pS () leads to Cross cross decoupling path transmission function P₁₂The output signal y of (s)_p12S () subtracts each other and obtains signal u_1p(s), i.e. u_1p(s)=u₁(s)- y_p12(s)；

C4：By signal u_1pS () acts on controlled device G₁₁S () obtains its output valve y₁₁(s)；By signal u_1pS () acts on controlled Object cross aisle transmission function G₂₁S () obtains its output valve y₂₁(s)；So as to realize to controlled device G₁₁(s) and G₂₁(s) Uneoupled control adds SPC, while realizing to network delay τ₁And τ₂Compensation with control；

The step of mode D, includes：

D1：Sensor S2 nodes work in time type of drive, and its trigger signal is cycle h₂Sampled signal；

D2：After sensor S2 nodes are triggered, to controlled device G₂₂The output signal y of (s)₂₂(s) and controlled device cross aisle Transmission function G₂₁The output signal y of (s)₂₁S () is sampled, and calculate the system output signal y of close loop control circuit 2₂ (s), and y₂(s)=y₂₂(s)+y₂₁(s)；

D3：By feedback signal y₂(s), by the feedback network path of close loop control circuit 2 to controller C node-node transmissions, feedback letter Number y₂S () will experience network transfer delay τ₄Afterwards, controller C nodes are got to；

The step of mode E, includes：

E1：Controller C nodes work in event driven manner, by feedback signal y₂S () is triggered；

E2：In controller C nodes, by the system Setting signal x of close loop control circuit 2₂(s), with feedback signal y₂(s) phase adduction After subtracting each other, signal e is obtained₂(s), i.e. e₂(s)=x₂(s)+y₂(s)-y₂(s)=x₂(s)；

E3：To e₂S () implements Internal Model Control Algorithm C_2IMCS (), obtains IMC signals u₂(s)；

E4：By IMC signals u₂S feedforward network path that () passes through close loop control circuit 2Unit is to decoupling actuator DA2 nodes Transmission, u₂S () will experience network transfer delay τ₃Afterwards, get to decouple actuator DA2 nodes；

The step of mode F, includes：

F1：Decoupling actuator DA2 nodes work in event driven manner, by IMC signals u₂S () is triggered；

F2：By IMC signals u₂S () decouples the output signal u of actuator DA1 nodes with close loop control circuit 1 is come from_1pS () leads to Cross cross decoupling path transmission function P₂₁The output signal y of (s)_p21S () subtracts each other and obtains signal u_2p(s), i.e. u_2p(s)=u₂(s)- y_p21(s)；

F3：By signal u_2pS () acts on controlled device G₂₂S () obtains its output valve y₂₂(s)；By signal u_2pS () acts on controlled Object cross aisle transmission function G₁₂S () obtains its output valve y₁₂(s)；So as to realize to controlled device G₂₂(s) and G₁₂(s) Uneoupled control adds IMC, while realizing to network delay τ₃And τ₄Compensation with control.

2. method according to claim 1, it is characterised in that：From TITO-NCS structures, realize system not comprising control The predict-compensate model of all-network time delay in loop 1 and control loop 2, so as to exempt to network delay τ between node₁And τ₂, And τ₃And τ₄Measurement, estimate or recognize, exempt the requirement synchronous to node clock signal.

3. method according to claim 1, it is characterised in that：Control loop 1 in TITO-NDCS uses SPC, due to from Realized in TITO-NDCS structures and specific controller C₁S the selection of () control strategy is unrelated, thus can be not only used for being controlled using conventional The TITO-NDCS of system, also can be used for using Based Intelligent Control or the TITO-NDCS using complex control strategy.

4. method according to claim 1, it is characterised in that：Control loop 2 in TITO-NDCS uses IMC, its internal model Controller C_2IMCS the adjustable parameter only one of which parameter of (), the regulation of its parameter is simple with selection, and explicit physical meaning；Adopt Stability, tracking performance and the interference free performance of system can be not only improved with IMC, but also the benefit to network delay can be realized Repay and IMC.