CN108665094A - Data-driven copper plate and strip casting-continuous rolling optimization scheduling method - Google Patents

Abstract

Establishing a layer of sensing model on a high-precision copper strip product production line; forming task requirements of a high-precision copper strip production line according to a production order, and establishing a high-precision copper strip casting-continuous rolling whole-flow operation scheduling model considering the production capacity of equipment, production process rules, product requirements and other constraints in the order through organic fusion of physical information of the high-precision copper strip; and designing a multi-objective intelligent optimization algorithm based on the plant root growth model to solve the operation scheduling model. The invention provides a fusion optimization scheduling model and an optimization method for a high-precision copper strip casting-tandem rolling production line, so that the production line can accurately control and remotely cooperate to integrate autonomy on the aspects of data acquisition, equipment state, material tracking, field scheduling and the like, the precision of a high-precision copper strip product and the overall utilization rate of equipment on the production line are improved, and the production cost is reduced.

Description

A data-driven optimal scheduling method for copper strip casting-continuous rolling

technical field

The invention relates to the field of information material fusion optimization scheduling method in high-precision copper strip casting-continuous rolling production, and more specifically, relates to a data-driven copper strip casting-continuous rolling optimization scheduling method.

Background technique

Copper strip products are high-end industrial products, and the production process is restricted by random and uncertain factors. Most of their orders present the characteristics of multiple varieties, small batches, and real-time requirements. This has great impact on the processing flexibility and intelligent scheduling of the copper strip production line. All made high demands. With the improvement of the equipment level of the copper strip production line, it is easier to obtain physical process information such as mass production and real-time status of equipment. However, due to the low degree of intelligence in the management and scheduling of copper strip production enterprises, the production line lacks precise control and remote collaboration comprehensive autonomy in terms of data collection, equipment status, material tracking, and on-site scheduling, so product accuracy, production costs, and efficiency It is difficult to meet the standards and consumes a lot of manpower, equipment resources and energy. It can be seen that in the production process of copper strip products, in the face of a large number of diverse and time-varying physical process information, a heterogeneous interconnected information-physical fusion architecture that runs through the underlying reliable perception model and the upper-level optimal scheduling method is needed.

The cyber-physical fusion scheduling optimization model established for high-precision copper plate production is a multi-agent spatio-temporal structure formed by interacting links or elements that reflects certain complexity, and its scheduling optimization objects are mostly large-scale, nonlinear, multi- Features such as goals, multi-stages, and large amount of heterogeneous information. Traditional dynamic programming algorithms can be used for small system models with few dimensions, but the problem of "curse of dimensionality" will also appear when facing multi-objective tasks. If the multi-objective is transformed into a single-objective solution strategy, the difference in weight values between the objectives and the mutual acceptance and anisotropy between different objectives cannot be ruled out, and the results are difficult to compare.

Contents of the invention

The invention overcomes the technical defect of the existing planning algorithm, and provides a new data-driven copper strip casting-continuous rolling optimization scheduling method. The invention provides a scheduling problem caused by the lack of precise control and remote cooperation comprehensive autonomy in the high-precision copper strip casting-continuous rolling production line, avoiding resource idleness and service collaborative optimization. The growth behavior improves the efficiency of the entire production line equipment.

In order to solve the problems of the technologies described above, the technical solution of the present invention is as follows:

A data-driven copper strip casting-continuous rolling optimization scheduling method, comprising the following steps:

S1: Set up a wireless sensor network on the copper strip production line. The wireless sensor network senses the physical environment around the copper strip production line and obtains perception information. The perception information includes the physical attributes of the product and the multimedia information of the equipment;

S2: Set the plant root growth model based on the perceived information of the copper strip production line;

S3: Through the TCP/IP network transmission protocol and socket data transmission protocol, the sensors of the copper strip production line are related to each other and the information transmission between the upper and lower layers of the system is realized;

S4: Combining knowledge discovery theory based on data mining and knowledge application theory based on data fusion, identify and process physical information and sensory data on the production line, perform data identification on sensory information on the production line, and remove redundancy in sensory information information;

S5: setting a constraint model, the constraint model is used to adjust the production plan;

S6: Solve the constraint model through the multi-objective optimization algorithm based on the plant root growth model to obtain the optimal solution of the production plan;

S7: Update the production plan according to the optimal solution, and execute the updated production plan.

In the present invention, biological experiments have proved that the morphogen concentration information that determines plant cell division and root growth is not given to cells in advance, but the cell system has received it from its surrounding environment, that is, the degree of soil fertility. According to this information, the plant roots show obvious trophic characteristics. The physical environment of the copper strip production line is regarded as soil, which contains various types of logic sensors and physical sensors for receiving various data. According to the modeling of the differentiation process of plant rhizome individuals, the external physical environment information perceived by the morpheme concentration is effectively organized and expressed, and passed to the upper layer to provide a unified application interface for application developers.

In a preferred scheme, the plant root growth model is represented by the following formula:

In the formula, the assumption is that there are n sensors on the production line, S _i = (S ₁ , S ₂ , ..., S _n ) means a specific sensor, E _i = (E ₁ , E ₂ , ..., E _n ) Indicates the perceptual information of the corresponding sensor; the f(*) is the objective function of the perceptual model.

In this preferred solution, the morphogen concentration of each growth point is determined by the relative position of each point and the environmental information of the position, which is the case for the generation mechanism of morphogen concentration in real plant cells. Therefore, n growth points correspond to n morpheme concentration values, and the concentration values will change each time new root hairs are produced.

In a preferred solution, said S4 includes the following sub-processes:

S4.1: Identify the outlier data in the perception information by selecting the cluster analysis algorithm or the gray clustering algorithm, and correct the outlier data. The outlier data will be stored separately after entering the database, and the outlier data will not be deleted;

S4.2: For the missing data collected, the linear interpolation method of recent stage data is used to fill in the gaps;

S4.3: Perform dimensionality reduction processing on the perception data through the nonlinear data transformation matrix;

S4.4: Discretize the continuous attribute values in the perception data, and generalize the perception data to a higher level through the concept hierarchy tree.

In a preferred solution, said S6 includes the following processes:

S6.1: According to the plant root growth model, initialize the seeds, generate a repeatable natural linked list L and a non-repeated natural number linked list I, thereby forming the seeds to be grown, the initial number of seeds is 2, the initial length of each rhizome is 1, and the growth point The threshold of the distance between them is 1;

S6.2: Calculate the objective function value, and modify the objective function value using constraints; perform non-dominated sorting on all individuals, and the fitness of each solution is the number of non-dominated layers;

S6.3: Select the growth point according to the non-dominated sorting method and the crowding distance, and the number of growth points selected for splitting is N; the value range of N is [2,64];

S6.4: Split the growth point by the single-point crossover method and the partial matching crossover method;

S6.5: Convert the non-repetitive natural number sequence I in the individual into a real number sequence Θ;

Transform the non-repeating natural number sequence I=(i ₁ ,i ₂ ,…,i _d ,…,i _n ) of each individual into an intermediate sequence Ψ=(ψ ₁ ,ψ ₂ ,…,ψ _d ,…,ψ _n ),Calculated as follows:

ψ _d =i _d -1

Transform the intermediate sequence into a real number sequence Θ=(θ ₁ ,θ ₂ ,…,θ _d ,…,θ _n ), the calculation formula is as follows:

θ _d =n-ψ _d +rand

Wherein, Ψ represents the collection of position indexes in descending order of Θ; the rand represents a random real number, and the value range of rand is [0,1];

S6.6: Perform a growth operation on the real number sequence Θ, and convert the real number sequence Θ into a non-repeating natural number sequence I;

Arrange Θ=(θ ₁ ,θ ₂ ,…,θ _d ,…,θ _n ) in descending order to get the position index set Ψ=(ψ ₁ ,ψ ₂ ,…,ψ _d ,…,ψ _n ), Then get I=(i ₁ ,i ₂ ,…,i _d ,…,i _n ) according to the formula:

In the formula, said d represents the dth virtual production order;

S6.7: Perform a mutation operation on the non-repeating natural number sequence I;

S6.8: Use constraint conditions to calculate the modified objective function value for all individuals, and then perform non-dominated sorting on the individuals. If the number of individuals exceeds the preset value, it will be screened by the crowding distance algorithm. If the number of iterations reaches the maximum number of cycles M , proceed to step 6.9; otherwise, repeat steps 6.3 to 6.7; said M is a preset positive integer;

S6.9: Perform Pareto optimization, and output the Pareto set according to the priority order, that is, the optimal production plan.

In a preferred solution, the number of iterations is counted by the rule:

After all individuals are mutated, the objective function value is calculated, and non-dominated sorting is performed, which is recorded as one iteration.

In a preferred solution, the number of divisions for each growth point in S6.4 is 4, and the single-point crossover probability and partial matching crossover probability are 0.85.

In a preferred solution, said S5 includes the following sub-processes:

S5.1: Define the following two decision variables to characterize the production sequence and production equipment:

Among them, i, j=0,1,...,n; i, j=0 represents the virtual production order, indicating the start and end of production; l represents the equipment, l=1,2,...,m;

S5.2: Define the objective function;

Among them, define the production adjustment time:

Wherein, i≠j; i, j=1,2,...,n, said n represents a random positive number; l=1,2,...,m, said m represents a random positive number; t _ijl represents in In processing equipment l, the processing adjustment time between production orders i and j; said T ₁ (X) represents the production adjustment time;

Define production time:

Wherein, i=1,2,...,n; l=1,2,...,m; t _il represents the processing time of production order i in the processing equipment l; said T ₂ (Y) represents the production adjustment time

Define the total production time:

The total production time is equal to the sum of production adjustment time and production time, and the total production time is calculated by the following formula:

In the formula, the f ₁ (X, Y) represents the total production time;

S5.3: Define constraints.

In a preferred solution, said S5.3 includes the following sub-processes:

S5.3.1: There is one and only one order after order i in the device, expressed by the following formula:

S5.3.2: There is one and only one order before order j in the device, expressed by the following formula:

S5.3.3: An order can only be arranged in one device or not for production, expressed by the following formula:

S5.3.4: The production of each piece of equipment includes a virtual order represented by the following formula:

S5.3.5: The total amount of orders being produced does not exceed the production capacity of the equipment in the current production cycle, expressed by the following formula:

Among them, the said Indicates the upper limit of the production capacity of each device in the production cycle; Indicates the lower limit of the production capacity of each device in the production cycle.

In a preferred solution, said S3 includes the following processes:

S3.1: The application model creates a socket according to the IP address type, socket type and TCP protocol;

S3.2: The application model end binds the IP address and port number for the socket;

S3.3: The socket on the application model end listens to the port number request, and is ready to receive the connection from the perception model end at any time. At this time, the socket on the application model end has not been opened;

S3.4: create a socket on the perception model side;

S3.5: Open the socket on the perception model side, and try to connect to the server socket according to the IP address and port number of the application model side;

S3.6: The socket on the application model side receives the client socket request, opens it passively, and starts to receive the request from the perception model side until the perception model side returns the connection information. At this time, the socket enters the blocking state, and the accept method does not return until the perception model end returns the connection information, and starts to receive the next perception model end request;

S3.7: Perceive the successful connection of the model end, and send connection status information to the application model end;

S3.8: The accept method on the application model side returns, and the connection is successful;

S3.9: The perception model end writes information to the socket;

S3.10: Read information on the application model side;

S3.11: The perception model end is closed;

S3.12: The application model side is closed.

Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

The bionic intelligent algorithm based on the firefly algorithm of the present invention is easy to implement for the optimization model, has strong global search ability, fast convergence speed, high solution accuracy, improves the quality and efficiency of production plan preparation, and also improves the efficiency of the production line. Overall utilization of equipment.

Description of drawings

Fig. 1 is the flowchart of embodiment.

Fig. 2 is a schematic diagram of sensory data collection of the copper strip casting-continuous rolling production line in the embodiment.

Fig. 3 is a flow chart of the multi-objective intelligent algorithm of plant root growth behavior in the embodiment.

Fig. 4 is the Pareto frontier diagram of the multi-objective intelligent algorithm optimization result of plant root growth behavior in the embodiment.

Detailed ways

The accompanying drawings are for illustrative purposes only and cannot be construed as limiting the patent;

In order to better illustrate this embodiment, some parts in the drawings will be omitted, enlarged or reduced, and do not represent the size of the actual product;

For those skilled in the art, it is understandable that some well-known structures and descriptions thereof may be omitted in the drawings.

The technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

As shown in Figure 1, a data-driven copper strip casting-continuous rolling optimization scheduling method includes the following steps:

S1: The sensory data collection method in the production of high-precision copper strips is shown in Figure 2. The sensory data is mainly divided into two categories. The first category changes with the change of processing space and time, or requires real-time collection of change data for real-time control, such as melting temperature, ambient humidity, process, location, number, processing time, etc., these The data is mainly collected by wireless sensors; the second type does not change with time and space, such as the operator's position and operating time, the name and number of the equipment, the number and specifications of the product, etc. These data are mainly used RFID electronic tags for collection.

S2: Set the plant root growth model based on the perceived information of the copper strip production line;

The plant root growth model is represented by the following formula:

In the formula, the assumption is that there are n sensors on the production line, S _i = (S ₁ , S ₂ , ..., S _n ) means a specific sensor, E _i = (E ₁ , E ₂ , ..., E _n ) Represents the perception information of the corresponding sensor; f(*) is the objective function of the perception model.

S3: Use TCP/IP network transmission protocol and Socket data transmission protocol to realize the mutual correlation of sensing elements on the high-precision copper strip production line and the information transmission between the upper and lower layers of the system. The connection must be established before transmitting data, and released after the end.

S3 includes the following processes:

S3.1: The application model creates a socket according to the IP address type, socket type and TCP protocol;

S3.2: The application model end binds the IP address and port number for the socket;

S3.3: The socket on the application model end listens to the port number request, and is ready to receive the connection from the perception model end at any time. At this time, the socket on the application model end has not been opened;

S3.4: create a socket on the perception model side;

S3.5: Open the socket on the perception model side, and try to connect to the server socket according to the IP address and port number of the application model side;

S3.6: The socket on the application model side receives the client socket request, opens it passively, and starts to receive the request from the perception model side until the perception model side returns the connection information. At this time, the socket enters the blocking state, and the accept method does not return until the perception model end returns the connection information, and starts to receive the next perception model end request;

S3.7: Perceive the successful connection of the model end, and send connection status information to the application model end;

S3.8: The accept method on the application model side returns, and the connection is successful;

S3.9: The perception model end writes information to the socket;

S3.10: Read information on the application model side;

S3.11: The perception model end is closed;

S3.12: The application model side is closed.

S4: Combining knowledge discovery theory based on data mining and knowledge application theory based on data fusion, identify and process physical information and sensory data on the production line, perform data identification on sensory information on the production line, and remove redundancy in sensory information information;

S4 includes the following sub-processes:

S4.1: Identify the outlier data in the perception information by selecting the cluster analysis algorithm or the gray clustering algorithm, and correct the outlier data. The outlier data will be stored separately after entering the database, and the outlier data will not be deleted;

S4.2: For the missing data collected, the linear interpolation method of recent stage data is used to fill in the gaps;

S4.3: Perform dimensionality reduction processing on the perception data through the nonlinear data transformation matrix;

S4.4: Discretize the continuous attribute values in the perception data, and generalize the perception data to a higher level through the concept hierarchy tree.

S5: According to the production order, form the task requirements of the high-precision copper strip production line, and through the organic fusion of the high-precision copper strip physical information, establish a high-precision copper strip melting and casting that considers the constraints of equipment production capacity, production process regulations, and product requirements in the order. -Operation scheduling model for the whole process of continuous rolling. The scheduling model is n production orders, m equipment for processing, each order contains multiple processes, the process sequence of each workpiece and the processing time of different machines are determined, and various constraints are met during the processing process to optimize the goal The total production time is equal to the production adjustment time, the sum of the production time and the penalty for not being included in the plan is the smallest;

S5.1: Define the following two decision variables to characterize the production sequence and production equipment:

Among them, i, j=0,1,...,n; i, j=0 represents the virtual production order, indicating the start and end of production; l represents the equipment, l=1,2,...,m;

S5.2: Define the objective function;

Among them, define the production adjustment time:

Among them, i≠j; i, j=1, 2,..., n, n represents a random positive number; l=1, 2,..., m, m represents a random positive number; t _ijl represents in the processing equipment l, the production The processing adjustment time between orders i and j; T ₁ (X) represents the production adjustment time;

Define production time:

Among them, i=1,2,...,n; l=1,2,...,m; t _il represents the processing time of production order i in processing equipment l; T ₂ (Y) represents the production adjustment time

Define the total production time:

The total production time is equal to the sum of production adjustment time and production time, and the total production time is calculated by the following formula:

In the formula, f ₁ (X, Y) represents the total production time;

S5.3: Define constraints:

S5.3.1: There is one and only one order after order i in the device, expressed by the following formula:

S5.3.2: There is one and only one order before order j in the device, expressed by the following formula:

S5.3.3: An order can only be arranged in one device or not for production, expressed by the following formula:

S5.3.4: The production of each piece of equipment includes a virtual order represented by the following formula:

S5.3.5: The total amount of orders being produced does not exceed the production capacity of the equipment in the current production cycle, expressed by the following formula:

in, Indicates the upper limit of the production capacity of each device in the production cycle; Indicates the lower limit of the production capacity of each device in the production cycle.

S6: Solve the constraint model through the multi-objective optimization algorithm based on the plant root growth model to obtain the optimal solution of the production plan. The multi-objective intelligent algorithm flow chart of the plant root growth behavior is shown in Figure 3;

S6.1: According to the plant root growth model, initialize the seeds, generate a repeatable natural linked list L and a non-repeated natural number linked list I, thereby forming the seeds to be grown, the initial number of seeds is 2, the initial length of each rhizome is 1, and the growth point The threshold of the distance between them is 1;

S6.2: Calculate the objective function value, and modify the objective function value using constraints; perform non-dominated sorting on all individuals, and the fitness of each solution is the number of non-dominated layers;

S6.3: Select the growth point according to the non-dominated sorting method and the crowding distance, and the number of growth points selected for splitting is 4;

S6.4: Split the growth point by the single-point crossover method and the partial matching crossover method;

S6.5: Convert the non-repetitive natural number sequence I in the individual into a real number sequence Θ;

Transform the non-repeating natural number sequence I=(i ₁ ,i ₂ ,…,i _d ,…,i _n ) of each individual into an intermediate sequence Ψ=(ψ ₁ ,ψ ₂ ,…,ψ _d ,…,ψ _n ),Calculated as follows:

ψ _d =i _d -1

Transform the intermediate sequence into a real number sequence Θ=(θ ₁ ,θ ₂ ,…,θ _d ,…,θ _n ), the calculation formula is as follows:

θ _d =n-ψ _d +rand

Among them, Ψ represents the set of position indexes of Θ in descending order; rand represents a random real number, and the value range of rand is [0,1];

S6.6: Perform a growth operation on the real number sequence Θ, and convert the real number sequence Θ into a non-repeating natural number sequence I;

Arrange Θ=(θ ₁ ,θ ₂ ,…,θ _d ,…,θ _n ) in descending order to get the position index set Ψ=(ψ ₁ ,ψ ₂ ,…,ψ _d ,…,ψ _n ), Then get I=(i ₁ ,i ₂ ,…,i _d ,…,i _n ) according to the formula:

In the formula, d represents the d-th virtual production order;

S6.7: Perform a mutation operation on the non-repeating natural number sequence I;

S6.8: Use constraint conditions to calculate the modified objective function value for all individuals, and then perform non-dominated sorting on the individuals. If the number of individuals exceeds the preset value, it will be screened by the crowding distance algorithm. If the number of iterations reaches the maximum number of cycles of 1000 , proceed to step 6.9; otherwise, repeat steps 6.3 to 6.7;

S6.9: Perform Pareto optimization, and output the Pareto set according to the priority order, as shown in Figure 4, that is, a set of optimal scheduling schemes.

S7: Send the decision-making and optimal scheduling control results generated by human-computer interaction to the production line through the actuator network, and receive feedback.

The terms describing the positional relationship in the drawings are only for illustrative purposes and cannot be interpreted as limitations on this patent;

Apparently, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, rather than limiting the implementation of the present invention. For those of ordinary skill in the art, other changes or changes in different forms can be made on the basis of the above description. It is not necessary and impossible to exhaustively list all the implementation manners here. All modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (9)

1. a data-driven copper strip casting-continuous rolling optimization scheduling method, is characterized in that, comprises the following steps: S1: Set up a wireless sensor network on the copper strip production line. The wireless sensor network senses the physical environment around the copper strip production line and obtains perception information. The perception information includes the physical attributes of the product and the multimedia information of the equipment; S2: Set the plant root growth model based on the perceived information of the copper strip production line; S3: Through the TCP/IP network transmission protocol and socket data transmission protocol, the sensors of the copper strip production line are related to each other and the information transmission between the upper and lower layers of the system is realized; S4: Combining knowledge discovery theory based on data mining and knowledge application theory based on data fusion, identify and process physical information and sensory data on the production line, perform data identification on sensory information on the production line, and remove redundancy in sensory information information; S5: setting a constraint model, the constraint model is used to adjust the production plan; S6: Solve the constraint model through the multi-objective optimization algorithm based on the plant root growth model to obtain the optimal solution of the production plan; S7: Update the production plan according to the optimal solution, and execute the updated production plan. 2. copper strip melting casting-continuous rolling optimization scheduling method according to claim 1, is characterized in that, described plant root growth model is represented by following formula: In the formula, the assumption is that there are n sensors on the production line, S _i = (S ₁ , S ₂ , ..., S _n ) means a specific sensor, E _i = (E ₁ , E ₂ , ..., E _n ) Indicates the perceptual information of the corresponding sensor; the f(*) is the objective function of the perceptual model. 3. according to claim 1 and 2 described copper strip casting-continuous rolling optimization dispatching method, it is characterized in that, described S4 comprises following subflow process: S4.1: Identify the outlier data in the perception information by selecting the cluster analysis algorithm or the gray clustering algorithm, and correct the outlier data. The outlier data will be stored separately after entering the database, and the outlier data will not be deleted; S4.2: For the missing data collected, the linear interpolation method of recent stage data is used to fill in the gaps; S4.3: Perform dimensionality reduction processing on the perception data through the nonlinear data transformation matrix; S4.4: Discretize the continuous attribute values in the perception data, and generalize the perception data to a higher level through the concept hierarchy tree. 4. The copper strip casting-continuous rolling optimization scheduling method according to claim 3 is characterized in that, described S6 comprises the following processes: S6.1: According to the plant root growth model, initialize the seeds, generate a repeatable natural linked list L and a non-repeated natural number linked list I, thereby forming the seeds to be grown, the initial number of seeds is 2, the initial length of each rhizome is 1, and the growth point The threshold of the distance between them is 1; S6.2: Calculate the objective function value, and modify the objective function value using constraints; perform non-dominated sorting on all individuals, and the fitness of each solution is the number of non-dominated layers; S6.3: Select the growth point according to the non-dominated sorting method and the crowding distance, and the number of growth points selected for splitting is N; the value range of N is [2,64]; S6.4: Split the growth point by the single-point crossover method and the partial matching crossover method; S6.5: Convert the non-repetitive natural number sequence I in the individual into a real number sequence Θ; Transform the non-repeating natural number sequence I=(i ₁ ,i ₂ ,…,i _d ,…,i _n ) of each individual into an intermediate sequence Ψ=(ψ ₁ ,ψ ₂ ,…,ψ _d ,…,ψ _n ),Calculated as follows: ψ _d =i _d -1 Transform the intermediate sequence into a real number sequence Θ=(θ ₁ ,θ ₂ ,…,θ _d ,…,θ _n ), the calculation formula is as follows: θ _d =n-ψ _d +rand Wherein, Ψ represents the collection of position indexes in descending order of Θ; the rand represents a random real number, and the value range of rand is [0,1]; S6.6: Perform a growth operation on the real number sequence Θ, and convert the real number sequence Θ into a non-repeating natural number sequence I; Arrange Θ=(θ ₁ ,θ ₂ ,…,θ _d ,…,θ _n ) in descending order to get the position index set Ψ=(ψ ₁ ,ψ ₂ ,…,ψ _d ,…,ψ _n ), Then get I=(i ₁ ,i ₂ ,…,i _d ,…,i _n ) according to the formula: In the formula, said d represents the dth virtual production order; S6.7: Perform a mutation operation on the non-repeating natural number sequence I; S6.8: Use constraint conditions to calculate the modified objective function value for all individuals, and then perform non-dominated sorting on the individuals. If the number of individuals exceeds the preset value, it will be screened by the crowding distance algorithm. If the number of iterations reaches the maximum number of cycles M , proceed to step 6.9; otherwise, repeat steps 6.3 to 6.7; said M is a preset positive integer; S6.9: Perform Pareto optimization, and output the Pareto set according to the priority order, that is, the optimal production plan. 5. the copper strip casting-continuous rolling optimization dispatching method according to claim 4, is characterized in that, described number of iterations is counted by rule: After all individuals are mutated, the objective function value is calculated, and non-dominated sorting is performed, which is recorded as one iteration. 6. according to claim 4 or 5 described copper strip casting-continuous rolling optimal dispatch method, it is characterized in that, the number of divisions of each growing point in the described S6.4 is 4, and the single-point intersection probability and part The matching crossover probability is 0.85. 7. The copper strip casting-continuous rolling optimization scheduling method according to claim 6, characterized in that, said S5 comprises the following sub-processes: S5.1: Define the following two decision variables to characterize the production sequence and production equipment: Among them, i, j=0,1,...,n; i, j=0 represents the virtual production order, indicating the start and end of production; l represents the equipment, l=1,2,...,m; S5.2: Define the objective function; Among them, define the production adjustment time: Wherein, i≠j; i, j=1,2,...,n, said n represents a random positive number; l=1,2,...,m, said m represents a random positive number; t _ijl represents in In processing equipment l, the processing adjustment time between production orders i and j; said T ₁ (X) represents the production adjustment time; Define production time: Wherein, i=1,2,...,n; l=1,2,...,m; t _il represents the processing time of production order i in the processing equipment l; said T ₂ (Y) represents the production adjustment time Define the total production time: The total production time is equal to the sum of production adjustment time and production time, and the total production time is calculated by the following formula: In the formula, the f ₁ (X, Y) represents the total production time; S5.3: Define constraints. 8. The copper strip casting-continuous rolling optimization scheduling method according to claim 7, characterized in that, said S5.3 includes the following sub-processes: S5.3.1: There is one and only one order after order i in the device, expressed by the following formula: S5.3.2: There is one and only one order before order j in the device, expressed by the following formula: S5.3.3: An order can only be arranged in one device or not for production, expressed by the following formula: S5.3.4: The production of each piece of equipment includes a virtual order represented by the following formula: S5.3.5: The total amount of orders being produced does not exceed the production capacity of the equipment in the current production cycle, expressed by the following formula: Among them, the said Indicates the upper limit of the production capacity of each device in the production cycle; Indicates the lower limit of the production capacity of each device in the production cycle. 9. according to claim 1, 2, 4, 5, 7 or 8 described copper strip casting-continuous rolling optimization scheduling method, it is characterized in that, described S3 comprises the following flow process: S3.1: The application model creates a socket according to the IP address type, socket type and TCP protocol; S3.2: The application model end binds the IP address and port number for the socket; S3.3: The socket on the application model end listens to the port number request, and is ready to receive the connection from the perception model end at any time. At this time, the socket on the application model end has not been opened; S3.4: create a socket on the perception model side; S3.5: Open the socket on the perception model side, and try to connect to the server socket according to the IP address and port number of the application model side; S3.6: The socket on the application model side receives the client socket request, opens it passively, and starts to receive the request from the perception model side until the perception model side returns the connection information. At this time, the socket enters the blocking state, and the accept method does not return until the perception model end returns the connection information, and starts to receive the next perception model end request; S3.7: Perceive the successful connection of the model end, and send connection status information to the application model end; S3.8: The accept method on the application model side returns, and the connection is successful; S3.9: The perception model end writes information to the socket; S3.10: Read information on the application model side; S3.11: The perception model end is closed; S3.12: The application model side is closed.