Immune memory learning control based wet coagulation bath temperature control process for carbon fiber precursor

1. Immunological memory-based learning control of the temperature of the coagulating bath silk aqueous method carbon fiber of the control process, the process line as polyacrylonitrile stock solution through the spinneret extrusion into the coagulating bath water tank, the temperature of the coagulating bath water in liquid by the temperature detecting element and feedback to the controller for real-time detection, controller according to the set point input, output feedback control quantity and the current and historical data to calculate the current control quantity, and outputs to the controlled object, the coagulating bath is heated, finally, the actual temperature reaches the set temperature of the coagulating bath, which is characterized in that: said controller based on immunological memory learning intelligent temperature controller; the immunological memory-based learning intelligent temperature controller consists of iterative learning controller, iterative learning storage unit, immune recognition module, immune response module and immune memory module.

2. Immunological memory-based carbon fiber of the learning control of the temperature of the coagulating bath silk aqueous method control process according to Claim 1, characterized in that the controlled object as the coagulating bath coagulating bath water tank and steam heating equipment; the coagulating bath steam heating equipment according to the control quantity control steam valve opening, so that a proper amount of high-temperature steam enters into the solidification of pipeline bath water trough , steam through the pipe wall and the liquid heat exchange occurs, the increase of temperature of the liquid; the coagulating bath steam heating equipment in coagulating bath temperature is lower than the set temperature and deviation reaches the setting temperature of the work to begin when the δ %, until the coagulating bath temperature reaches the set temperature of the work is stopped, this process is known as between one working of the heat exchanger; wherein the δ ∈ [0,100] the process precision coefficient, selected according to the actual process requirements.

3. Immunological memory-based carbon fiber of the learning control of the temperature of the coagulating bath silk aqueous method control process according to Claim 1, characterized in that said iterative learning controller and the iterative learning storage unit according to the establishment of iterative learning control algorithm, the learning controller states the iteration of the heat exchanger a working as an iterative process, according to a certain learning law, through the continuous iteration in the historical data of the learning control system, in order to continually improve the control effect; the stated study the law is any iterative learning law.

4. Immunological memory-based carbon fiber of the learning control of the temperature of the coagulating bath silk aqueous method control process according to Claim 1, characterized in that said immune recognition module simulating human specific immune process identification of the immune system in the mechanism of the specific antigen, in the control system the interference as the antigen, the antigen into the specific identification; the antigen Ag_i defined as:

Agi={r,u,e,y},i=1,...,Nummaxim,

Wherein

R= {r (t), r (t-1), ..., r(t-n)};

R (t) t time system is input, (t-1) r is t-1 time system input, is r(t-n) t-n time system input;

U= {u (t), u (t-1), ..., u(t-n)};

U (t) t output of the controller is at all times, (t-1) u is t-1 output of the controller at all times, at all times is u(t-n) t-n output of the controller;

E= {e (t), e (t-1), ..., e(t-n)};

E (t) t the system error is time, e (t-1) is t-1 system error at all times, the system error is e(t-n) t-n time;

Y= {y (t), y (t-1), ..., y(t-n)};

Y (t) is the system output at all times t, y (t-1) is t-1 time system output, the system output is y(t-n) t-n time;

The length of n-antigen;

-Immunological memory module the maximum storage capacity;

I-antigen number, i is a constant,

The identifying means according to the specificity of the at the moment when the interference r, u, the value of e and y of identifying the same Ag_i;

Modular simulation of the immune response in the human specific immune process mechanism of the immune response, to the antigen with different situation, initial, appear or reoccurs, to make different response mechanism, that is, the initial response and once again the response, and output with antigen matched antibody; the antibody Ab_i defined as:

Abi=uiim,i=1,...,Nummaxim

Wherein

uiim={uiim(1),...,uiim(m)};

Is directed against antigen Ag_i time after the emergence of the immune control section m;

The length of the m-antibody;

The antigen of the matching relationship with the antibody is defined as:

{Ag_i, Ab_i};

The immunological memory module specific immune simulating human immunological memory mechanism in the process, to the antigen specificity and its matching of the antibody in accordance with the stored matching relationship, and for immune recognition module and module to visit immune response, query and reading; said immune memory module is a linear memory, when the memory remaining capacity is not sufficient to store the new data, to automatically delete access probability according to the searched access probability the most low data, and then store new data.

5. Immunological memory-based carbon fiber of the learning control of the temperature of the coagulating bath silk aqueous method control process according to Claim 4, characterized in that the intelligent of immunological memory-based learning algorithm flow of a temperature controller is as follows:

1) specific identification:

A. by the immunological recognition module, and the real-time monitoring the r (t), e (t) and u (t) three signal, and determining whether the present time of the real-time monitoring of whether interference occurs; when the following conditions are fulfilled:

e (t) ≥ (| u (t)-u (t-1) | +r (t)) ·ε_im,

At this moment, in other words, that presence of interference, and to initiate immune response module immunological memory module;

Wherein ε_im is the interference sensitivity coefficient, ε_im immunological recognition too large for the smaller is the error does not respond to, high control precision; ε_im is too small it will produce a large number of antigen, the computation speed is relatively slow; ε_im can be many attempts the practical application, appropriate selection of;

The interference moment of the b. r, u, and e as the a antigen y Ag_i, and to identify the specificity of the antigen, i.e., the immunological memory module to search in the search, if the search to the same antigen the antigen number recording is obtained, and that the antigen invasion again, if the lookup is not the as a new antigen given new and that the number of the initial antigen invasion;

The c. after antigen-specific recognition of the antigen number, and that initial or invasion again, the immune response is sent to the module, to initiate immune response module;

2) response:

To initiate immune response module response algorithm divided into primary response and secondary response two situations, antigen initial invasion of the initial response, again re-invasion of the antigen response;

The initial acknowledgement includes the following steps:

(1) by using the iterative learning controller of the iterative learning control algorithm and its learning the law is right the antigen be caused by the output error of the, controller output corresponding to the controlled quantity and ultimately reduce the output error of the output error tends to zero; at the same time, the error elimination process output controller in the control amount for u matched antibody with the antigen Wherein u= {u (t), u (t+ 1), ..., u(t+m)};

(2) the elimination of error, the initial the antigen Ag_i and its matching antibody Ab_i in accordance with the matching relations {Ag_i, Ab_i} storage to immunological memory module;

The again acknowledgement includes the following steps:

(1) according to the immunological recognition module memory module found matching relationship between antigen of the antibody, the antigen matched reading the recording of the antibody, i.e., obtain controller at a time against the antigen in the immune control quantity at the time of

(2) the previous step (1) with the antibodies obtained in real-time iterative learning controller with the output of the, formed and directed against the antigen, and the control output, and according to the learning law further update, the output to the controlled object;

The defined as the combination of:

uk′(t)=α[!CenterDot!]uk(t)+(1-α)[!CenterDot!]uiim

Wherein

K the lower header of section k time data is expressed as the data in the iterative process, k+ 1 expressed as the subsection k + 1 times the data in the iterative process;

u_k (t)-iterative learning controller section k iteration t time in the course of output of the control quantity;

Α-smoothing parameter, α ∈ [0, 1];

K iteration-t time in the course of the immune response module output;

The updating is defined as:

Wherein

K iteration-t time in the course of the immune response module output;

u_k+1 (t)-section k + 1 iteration t time in the course of output of the controller;

Φ, Ψ, the parameter Γ-PID-shaped mold;

(3) the elimination of antigen, the updated control amount u_k+1 (t) recorded as the antigen Ag_i matching antibody Ab_i, according to the matching relationship and {Ag_i, Ab_i} storage to immune memory module, the replacement before the updating Ag_i matching of the antibody.