Continuous dyeing temperature zone multi-point cooperative control method

Abstract

The application provides a continuous dyeing temperature zone multi-point cooperative control method, comprising the following steps: collecting the measured temperature and the set temperature of each temperature zone of a continuous dyeing machine, calculating the temperature zone deviation, and encoding the heating power correction amount arranged in the order of the temperature zones into an initial population; determining a scaling factor according to the temperature zone deviation and the number of cross generations, combining a difference vector to generate a variation individual; dividing the continuous temperature zone segments according to the positive and negative changes of the adjacent dimension power correction amount difference value, determining the cross probability according to the segment deviation mean value, and exchanging the target individual and the variation individual to generate a test individual in the segment unit to maintain the cooperative correction relationship of the adjacent temperature zones; inputting the candidate power correction amount, the measured temperature and the current basic heating power into a temperature prediction model to obtain a predicted temperature and calculate the fitness, and reserving the individuals by selection; when the fitness is stagnant, the worst individual is disturbed to maintain the population diversity, and after the iteration is completed, the optimal individual is decoded and sent to an actuator to complete temperature adjustment.

Description

Technical Field

[0001] This application belongs to the field of collaborative control, and in particular relates to a multi-point collaborative control method for continuous staining temperature zones. Background Technology

[0002] Continuous dyeing machines typically contain multiple heating units and temperature sensors. The temperature state of each zone is affected not only by its own heating power but also by the heat transfer from adjacent zones due to continuous fabric transport and airflow. The control system calculates the temperature zone deviation between the measured temperature and the set temperature based on the measured temperature collected by the temperature sensors and coordinates the heating power of each zone. Existing multi-point coordinated control methods calculate the power control quantities of multiple temperature zones by establishing a temperature prediction model or using decoupling algorithms, and then send commands to the heating actuators of each zone to maintain the overall temperature field balance. However, existing temperature optimization control strategies based on conventional heuristic algorithms generally suffer from drawbacks when facing high-dimensional and complex temperature zone environments. These drawbacks include a lack of specificity in cross-variation operations, a tendency to get trapped in local optima during the optimization process, and weak ability to coordinate and adjust local continuous temperature zones. They cannot adaptively optimize power based on real-time dynamic deviations and adjacent temperature differences between temperature zones, making it difficult to meet the requirements of modern continuous dyeing machines for high-precision, low-fluctuation, and global dynamic coordinated control of multiple temperature zones. Summary of the Invention

[0003] To determine the heating power correction amount that meets the fitness requirements under the current iterative conditions and ensure the stable operation of the dyeing process, according to one aspect of the present invention, a multi-point coordinated control method for a continuous dyeing temperature zone is provided, the method comprising:

[0004] Based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, the temperature zone deviation is obtained. The heating power correction of each temperature zone is encoded into individuals according to the temperature zone order to construct an initial population. Individuals to be crossovered are selected from the current population as target individuals. The scaling factor is calculated based on the temperature zone deviation and the number of crossover adoptions in previous generations. Combined with the difference vector, mutant individuals are generated. Individuals are divided into several continuous temperature zone segments based on the positive and negative changes in the difference of heating power correction between adjacent dimensions. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. Experimental individuals are obtained by exchanging segments between the target individual and the mutated individual. The individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone are input into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones, the fitness value is calculated to select the best individual to retain. If the fitness value does not improve, a perturbation individual is generated to replace the worst individual in the population. When the iteration reaches the termination condition, the optimal individual is decoded into a heating power correction amount and sent to the actuator.

[0005] According to another aspect of the present invention, a multi-point coordinated control system for continuous staining temperature zones is provided, the system comprising the following modules: The generation module is used to obtain the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, encode the heating power correction of each temperature zone into individuals according to the temperature zone order to construct the initial population, select individuals to be crossovered from the current population as target individuals, calculate the scaling factor based on the temperature zone deviation and the number of crossover adoptions in previous generations, and generate mutant individuals by combining the difference vector. The determination module is used to divide an individual into several continuous temperature zone segments based on the positive and negative changes in the difference of heating power correction between adjacent dimensions within the individual. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. The test individual is obtained by exchanging segments between the target individual and the mutated individual. The replacement module is used to input the individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones, the fitness value is calculated to select the best individual to retain. If the fitness value does not improve, a perturbation individual is generated to replace the worst individual in the population. When the iteration reaches the termination condition, the optimal individual is decoded into a heating power correction amount and sent to the actuator.

[0006] This invention calculates a scaling factor based on temperature zone deviation and the number of crossover adoptions in each iteration to rationally control the variable temperature length. It divides continuous temperature zones into segments based on the change in the difference between adjacent temperature zone corrections, and determines the crossover probability based on the average segment deviation, performing crossover exchanges on a segment-by-segment basis. This strategy fully considers the thermal coupling effect between adjacent temperature zones, preserving excellent local collaborative control information. Fitness evaluation is performed by combining the measured temperature of each temperature zone, the current base heating power, and the candidate heating power correction input temperature prediction model. When optimization stalls, perturbed individuals are used to replace the worst individual, maintaining population diversity. This yields the heating power correction that meets the fitness requirements under the current iteration conditions, which helps improve the coordination and stability of multi-temperature zone temperature control and reduce temperature fluctuations in the dyeing process. Attached Figure Description

[0007] Figure 1 A flowchart of a multi-point coordinated control method for continuous staining temperature zones; Figure 2 This is a graph showing the changes in optimal fitness of a population due to stagnation and perturbation. Detailed Implementation

[0008] In this application, the terms "exemplarily" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplarily" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner.

[0009] In this application, a method for multi-point coordinated control of continuous staining temperature zones is provided, such as... Figure 1 As shown, the method includes: S1. Based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, the temperature zone deviation is obtained. The heating power correction of each temperature zone is encoded into individuals in the order of temperature zone to construct an initial population. Individuals to be crossovered are selected from the current population as target individuals. The scaling factor is calculated based on the temperature zone deviation and the number of crossover adoptions in previous generations. Combined with the difference vector, mutant individuals are generated.

[0010] The measured temperature data of each temperature zone is collected in real time by platinum resistance temperature sensors installed inside each independent heating zone of the continuous dyeing machine, and transmitted to the programmable logic controller via an analog-to-digital converter. The absolute value of the difference between the measured temperature of each temperature zone and the corresponding set temperature is calculated as the current temperature zone deviation of each temperature zone, which is used to characterize the degree to which each temperature zone deviates from the target temperature.

[0011] The heating power correction values ​​for all temperature zones are arranged in the order of the physical temperature zones from fabric entry to exit in the continuous dyeing machine, forming a multi-dimensional vector as an independent individual in the evolutionary algorithm. Each dimension represents the heating power correction value for the corresponding temperature zone. Multiple such multi-dimensional vectors are randomly generated within the allowable upper and lower limits of the heating power correction values ​​for each temperature zone, thus constructing an initial population of fixed size. In each evolutionary iteration, individuals to be crossovered are sequentially selected from the current population as target individuals, and corresponding mutant individuals and experimental individuals are generated around the target individuals.

[0012] During the evolutionary iteration process, a mutation scaling factor is calculated for each dimension of an individual. The calculation method is as follows: obtain the fitness value of each individual in the current population, sort all individuals in ascending order of fitness value, and select the individual at the sorting index of half the population size as the basis vector; for any dimension of an individual, when the sum of all temperature zone deviations is greater than a preset minimum positive number, divide the temperature zone deviation corresponding to that dimension by the sum of all temperature zone deviations to obtain the basic scaling factor of that dimension; when the sum of all temperature zone deviations is less than or equal to the preset minimum positive number, set the basic scaling factor of each dimension to a preset equal value, or stop the current round of mutation optimization and keep the current heating power correction unchanged; before the initial iteration, initialize the historical adoption count of each dimension to zero; read the historical adoption count of that dimension that was adopted and retained during the historical iteration process, divide the preset constant by the sum of the historical adoption count and the preset ratio constant to obtain the bias gain, and then multiply the basic scaling factor by the bias gain to obtain the scaling factor of that dimension.

[0013] Two individuals from the current population, different from the basis vector, are randomly selected. The corresponding dimension values ​​of the two individuals are subtracted to obtain a difference vector. The difference vector is then multiplied by the scaling factors of each dimension calculated above, and then added to the basis vector to generate a mutated individual containing preliminary power correction information for each temperature zone. If any mutated value in any dimension exceeds the closed-interval constraint range of the heating power correction amount for that temperature zone, the mutated value in that dimension is truncated to the corresponding closed-interval boundary value to ensure that the mutated individual meets the physical control constraints of the actuator.

[0014] In some implementations, the step of obtaining the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous staining machine, and encoding the heating power correction amount of each temperature zone into individuals to construct the initial population in the order of the temperature zones includes: The target operating temperature of each temperature zone preset by the control system is obtained as the set temperature, and the real-time measurement value of the temperature sensor of each temperature zone is collected as the measured temperature. The absolute value of the difference between the measured temperature and the set temperature in each temperature zone is calculated as the temperature zone deviation; Determine the closed-range constraint range formed by the maximum allowable decrease threshold and the maximum increase threshold of the actuator in each temperature zone. The closed-range constraint range ensures that the sum of the current basic heating power of each temperature zone and any heating power correction amount is not less than zero and does not exceed the rated power limit of the corresponding actuator. Within the closed interval constraint range, a number of real numbers equal to the total number of temperature zones are randomly generated. The generated real numbers are arranged into row vectors according to the physical arrangement order of the temperature zones to construct an individual. Each dimension of the individual corresponds to the heating power correction amount of the corresponding temperature zone. The individuals are generated repeatedly until the total number of individuals reaches the preset population size, thus constructing the initial population.

[0015] Taking a continuous dyeing machine with 8 temperature zones (N=8 in total) as an example, the PT100 temperature sensor readings are read in real time via the communication bus. And obtain the target operating temperature from the preset process file. Set temperature The values ​​are [100, 105, 110, 115, 120, 120, 125, 125], representing the actual measured temperatures. The values ​​are [98, 106, 108, 117, 119, 122, 123, 126] in sequence. These values ​​are obtained through absolute value operations. Calculate the deviation for each temperature zone to obtain the deviation vector E=[2,1,2,2,1,2,2,1].

[0016] Set a closed-range constraint range for the heating power correction amount, such as the maximum reduction threshold. kW, maximum increase threshold kW, meaning the value constraint interval is [-15.0, 15.0]. Eight real numbers are randomly generated within this interval, for example, [5.2, -1.1, 4.3, -2.5, 1.8, -3.2, 3.1, -1.5]. These eight real numbers are arranged in physical order from temperature zone 1 to temperature zone 8 to form a row vector, constituting an individual. At this point, the value of the i-th dimension of the vector corresponds to the heating power correction amount for the i-th temperature zone.

[0017] Repeat the row vector generation process described above until the total number of individuals generated reaches the pre-set population size, such as NP=50. Construct a 50x8 matrix as the initial population, where each row of the matrix corresponds to one individual.

[0018] In some implementations, the step of calculating the scaling factor based on the temperature zone deviation and the number of cross-adoptions in each generation, and then combining it with the difference vector to generate variant individuals, includes: Obtain the fitness values ​​of each individual in the current population and sort them in ascending order. Select the individual whose sorting position is half the population size as the basis vector. For any dimension, when the sum of all temperature zone deviations is greater than a preset minimum positive number, the ratio of the temperature zone deviation corresponding to that dimension to the sum of the temperature zone deviations is used as the basic scaling factor. When the sum of all temperature zone deviations is not greater than the preset minimum positive number, the basic scaling factor of each dimension is set to the preset equal value; Before the initial iteration, the historical adoption count of each dimension is initialized to zero, and the historical adoption count of that dimension that was adopted and retained during the historical iteration process is read as the cross-adoption count of each generation. The bias gain is determined according to the ratio of the preset constant to the sum of the historical adoption count and the preset ratio constant. The scaling factor of that dimension is obtained by multiplying the basic scaling factor by the bias gain. Randomly select two individuals from the current population that are different from the basis vector, and calculate the difference between the two in the same dimension as the difference vector; The difference vector is multiplied by the scaling factor of the corresponding dimension and then superimposed onto the basis vector to obtain the mutation value of each dimension. The mutation values ​​of each dimension are combined to generate mutated individuals. The mutation values ​​that exceed the closed interval constraint range are truncated at the boundary.

[0019] Obtain the fitness values ​​of all individuals in the current population of size 50 and sort them in ascending order. Select the individual at sort index 25 as the basis vector. For any dimension i of an individual, such as the first dimension, the aforementioned measured temperature zone deviation... And the sum of the deviations in each temperature zone Calculate the base scaling factor .

[0020] Read the number of times this dimension has been adopted and retained in historical iterations. Assuming that the number of historical adoptions in the first dimension is [number] when iterating to the current generation. And according to preset constants With the preset ratio constant Calculate the bias gain Multiplying the base scaling factor by the bias gain yields the scaling factor for that dimension. .

[0021] After calculating the scaling factors for each dimension, randomly select two individuals from the current 50 individuals that are different from the previous ones. individual and Suppose we extract individuals with indices 12 and 37. Taking the first dimension as an example, we calculate the difference between the two individuals in this dimension as the difference vector. If the power corrections for the two are 4.5 and 1.5 respectively, then the difference vector .

[0022] Multiply the difference vector by the scaling factor for that dimension, and then add it to the value of the basis vector in the corresponding dimension to obtain the mutation value. Calculate the variance values ​​for all 8 dimensions, and combine the variance values ​​of each dimension to generate a variant individual V=[2.385,...]. If the variance value of a certain dimension exceeds the constraint interval of [-15.0, 15.0]kW, for example, if the calculated value is 16.2kW, then the variance value of that dimension is truncated to 15.0kW; if the calculated value is -17.4kW, then it is truncated to -15.0kW.

[0023] S2, the individual is divided into several continuous temperature zone segments according to the positive and negative changes in the difference of heating power correction between adjacent dimensions in the individual. The segment crossover probability is determined based on the ratio of the average temperature zone deviation in each segment to the average deviation of the largest segment. The test individual is obtained by exchanging segments between the target individual and the mutated individual.

[0024] Iterate through each dimension of the current individual's multidimensional vector, subtracting the heating power correction of the previous dimension from the heating power correction of the next dimension to calculate the difference in heating power correction between adjacent dimensions. Record the spatial index positions where these differences change from positive to negative or vice versa. Use these index positions where the signs of the differences reverse as cutting points to sequentially truncate the entire multidimensional vector of the individual into multiple continuous temperature zone segments of varying lengths, ensuring that the power correction trend within each segment remains consistent. When the difference in heating power correction between adjacent dimensions is zero, do not use this zero-value difference point as a separate criterion for sign reversal, or use the sign of the previous non-zero difference in subsequent segmentation judgments.

[0025] For each consecutive temperature zone segment, extract the absolute values ​​of the temperature zone deviations of all temperature zones contained in that segment at the current time and calculate their arithmetic mean, which is taken as the mean temperature zone deviation of that segment. Compare the mean temperature zone deviations of all segments and select the maximum value as the maximum mean segment deviation. When the maximum mean segment deviation is greater than a preset minimum positive number, divide the mean temperature zone deviation of the current segment by the maximum mean segment deviation to obtain a proportionality coefficient between 0 and 1, and use this proportionality coefficient as the segment crossover probability of the corresponding segment; when the maximum mean segment deviation is less than or equal to the preset minimum positive number, set the segment crossover probability of each temperature zone segment to a preset equal probability or a preset low probability to avoid division by zero and to ensure that the crossover operation is executable.

[0026] When performing crossover operations, instead of using a single temperature zone dimension as the unit, a pre-defined continuous temperature zone segment is treated as a whole. A random number between 0 and 1 is generated and compared with the crossover probability of that segment. If the random number is less than or equal to the segment crossover probability, the values ​​of all dimensions corresponding to that segment in the experimental individual are replaced with the values ​​of the corresponding segment in the mutant individual. Otherwise, the values ​​of that segment in the target individual are retained and assigned to the experimental individual, thus completing the multi-point co-regulation experimental individual.

[0027] In some implementations, the individual is divided into several continuous temperature zone segments based on the positive and negative changes in the heating power correction difference between adjacent dimensions. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. Experimental individuals are obtained by exchanging segments between the target individual and the mutated individual. This includes: Calculate the difference in heating power correction between adjacent dimensions in the target individual. When two adjacent differences are both non-zero and the sign is reversed, set a segment point at the intersection of the corresponding dimensions. When the difference is zero, it is not used as a basis for reversing the sign. The target individual is divided into several consecutive temperature zones based on all the aforementioned segmentation points; Calculate the average of all temperature zone deviations within each temperature zone segment as the segment deviation mean, and select the largest segment deviation mean as the maximum segment deviation mean. When the mean of the maximum segment deviation is greater than the preset minimum positive number, the mean of the segment deviation is divided by the mean of the maximum segment deviation to obtain the segment crossover probability corresponding to each temperature zone segment; when the mean of the maximum segment deviation is not greater than the preset minimum positive number, the segment crossover probability corresponding to each temperature zone segment is set to a preset equal probability or a preset low probability. For each temperature zone segment, a random number between 0 and 1 is generated. When the random number is less than the corresponding segment crossover probability, the value of the corresponding temperature zone segment in the mutated individual is replaced with the corresponding position in the target individual. Otherwise, the original value of the temperature zone segment in the target individual is retained. After completing the crossover operation of all temperature zone segments, the experimental individual is obtained.

[0028] Extract target individual vectors of length 8 Calculate the difference in heating power correction between adjacent dimensions. , , , , , , The sign of this sequence changes as [+,-,-,+,+,-,+]. and , and , and , and The sign is reversed at this point.

[0029] Based on these positive and negative boundaries, segmentation points are set to divide the target individual into continuous temperature zone segments: segment 1 contains dimensions [1,2], segment 2 contains [3,4], segment 3 contains [5,6], segment 4 contains [7], and segment 5 contains [8]. Combining the aforementioned actual temperature zone deviation vector E=[2,1,2,2,1,2,2,1], the mean deviation within each segment is calculated.

[0030] The mean deviation for segment 1 is 1.5, segment 2 is 2.0, segment 3 is 1.5, segment 4 is 2.0, and segment 5 is 1.0. The maximum mean segment deviation, MaxE = 2.0, is selected. The segment crossover probability CR for each temperature range segment is obtained by dividing the mean segment deviation by the maximum mean segment deviation, where segment 1's... Fragment 2 Fragment 5 =0.5.

[0031] For fragment 1, a random number rand=0.62 is generated within the interval [0,1]. Since 0.62 < 0.75, the value of the mutated individual corresponding to dimension [1,2] is replaced to the corresponding fragment position of the target individual. If the random number rand=0.8 generated for fragment 5 is greater than 0.75, then... If the original value of the 8th dimension of the target individual is retained, then the crossover operation of all 5 temperature range segments is completed to obtain the experimental individual. .

[0032] S3, input the individual, the measured temperature of each temperature zone and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Calculate the fitness value based on the predicted temperature, the set temperature and the target temperature difference between adjacent temperature zones to select the best individual. If the fitness value does not improve, generate a perturbation individual to replace the worst individual in the population. When the iteration reaches the termination condition, decode the best individual into a heating power correction amount and send it to the execution mechanism.

[0033] The continuous dyeing machine temperature prediction model is a multilayer perceptron neural network model pre-trained using historical heating power data, historical measured temperature data, and corresponding temperature response data. The basic heating power output of each temperature zone is read from the continuous dyeing machine controller, and a closed range of available power correction is determined based on the rated power limit of the actuators in each temperature zone. This ensures that the sum of the current basic heating power and the heating power correction is not less than zero and does not exceed the rated power limit of the corresponding actuator. The heating power correction for each temperature zone included in the test individual is added to the current basic heating power, and after power range limiting, candidate heating powers are obtained. These candidate heating powers and the current measured temperatures of each temperature zone are used as input features. Before being input into the neural network model, the candidate heating powers and the current measured temperatures of each temperature zone are normalized according to the normalization parameters saved during the model training phase. When the neural network model outputs a normalized temperature value, the predicted temperature of each temperature zone is obtained by restoring it according to the corresponding inverse normalization parameters. This predicted temperature is then input into the neural network model for forward propagation calculation, and the predicted temperature of each temperature zone after a control delay is output. The prediction deviation is obtained by subtracting the set target temperature from the predicted temperature of each temperature zone. The sum of the squares of the prediction deviations of all temperature zones is used as the first evaluation item. At the same time, the sum of the squares of the deviations between the predicted temperature difference of adjacent temperature zones and the target temperature difference of the corresponding adjacent temperature zones is calculated as the second evaluation item. The fitness value is obtained by weighted summing of the above two items. The smaller the fitness value, the better the control effect of the individual.

[0034] By comparing the fitness values ​​of the experimental individual with those of the original target individual, if the fitness value of the experimental individual is better than that of the original target individual, the experimental individual replaces the original target individual in the next generation of the population; otherwise, the original target individual is retained in the next generation. A stagnation counter is set up to record the optimal fitness value in the population after each generation of evolution. If the optimal fitness value does not decrease for a specified number of generations, it is determined that the fitness value has not improved. At this time, the target dimension with the largest deviation between the predicted temperature and the set temperature is determined in the current optimal individual. A new real number is randomly generated within the closed interval constraint to reassign the target dimension. The values ​​of the other dimensions of the current optimal individual are kept unchanged to generate a perturbation individual. This perturbation individual replaces the worst individual with the largest fitness value in the population, thereby increasing population diversity and escaping local optima.

[0035] The algorithm stops iterating when the number of iterations reaches the pre-set maximum number of iterations, or when the fitness value of the best individual in the population is less than the allowable minimum error threshold. The individual with the best fitness value in the current population is extracted, and the data of each dimension of the multi-dimensional vector is decomposed into independent digital control signals according to the correspondence of physical temperature zones. These signals are then converted into power adjustment commands recognizable by the corresponding actuators via a control network. These power adjustment commands are then sent to the solid-state power regulators or solid-state relays in the corresponding temperature zones via an industrial fieldbus, changing the equivalent heat output of the corresponding temperature zones and completing multi-point coordinated temperature control.

[0036] In some implementations, the step of inputting the individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone, and calculating the fitness value based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones to select the best individuals, includes: The heating power correction for each temperature zone corresponding to an individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone are all input into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. The difference between the set temperatures of two adjacent temperature zones is determined as the target temperature difference for the corresponding adjacent temperature zones. The sum of squares of the differences between the predicted temperature and the set temperature in each temperature zone of an individual is calculated and multiplied by a first weighting coefficient to obtain the first evaluation item; The second evaluation term is obtained by calculating the sum of squared deviations between the predicted temperature difference between adjacent temperature zones and the target temperature difference between the corresponding adjacent temperature zones, and multiplying the sum of squared deviations by the second weighting coefficient. The fitness value of an individual is obtained by adding the first evaluation item to the second evaluation item. The fitness values ​​of the experimental individuals and the target individuals are calculated respectively. The individuals with lower fitness values ​​are compared and retained into the next generation of the population to achieve the selective retention of individuals. If the fitness value of the experimental individual is less than the fitness value of the target individual, then the temperature range segments that have been exchanged in the experimental individual are checked. For temperature range segments that are taken from the mutant individuals and win the selection and are retained, the historical adoption count of the corresponding dimension in the statistical pool is incremented by 1, which is used to correct the scaling factor of the corresponding dimension in the next generation.

[0037] The current individual's 8-dimensional heating power correction sequence, the 8-dimensional measured temperature sequence for each temperature zone, and the current basic heating power sequence for each temperature zone are all used as input features and fed into the input layer of the multilayer perceptron neural network of the continuous dyeing machine temperature prediction model. The output of the previous hidden layer is linearly weighted and summed using the corresponding weight matrix and bias vector, and then processed by the activation function to obtain the output of the current hidden layer. The calculation formula is as follows: ,in This represents the output feature of the l-th layer. This represents the weight matrix of the layer. This represents the output features of the (l-1)th layer. This represents the bias vector of this layer. This represents the activation function.

[0038] After forward inference, the predicted temperature for each temperature zone is output by the output layer. Calculate the predicted temperature and set temperature for each temperature zone. The sum of squares of the differences and multiplied by the first weighting factor Received the first evaluation item Calculate the sum of squared deviations between the predicted temperature difference of adjacent temperature zones and the target temperature difference of the corresponding adjacent temperature zones, and multiply it by the second weighting coefficient. The second evaluation item was obtained. The sum of the two values ​​above yields the fitness value of the current individual.

[0039] Assuming the fitness value of the experimental individuals is obtained target individuals The fitness values ​​of the two individuals are compared. Because the fitness value of the experimental individual is lower, it is retained for the next generation of the population. Simultaneously, the temperature range segments where exchange occurred within the experimental individual are examined. For segments taken entirely from the mutated individual and retained, the historical adoption count for the corresponding dimension is incremented by one. For example... , and This is used to correct the scaling factor of the corresponding dimension in the next generation.

[0040] In some implementations, generating a perturbed individual to replace the worst individual in the population when the fitness value does not improve includes: After each generation iteration, record the best individual with the smallest fitness value in the current population and its fitness value; Compare the fitness value of the current generation's best individual with the historical best fitness value. If there is no decrease, increment the number of generations without improvement by 1. When the unimproved algebra reaches a preset stagnation algebra threshold, the target dimension with the largest deviation between the predicted temperature and the set temperature in the current best individual is determined. Within the closed interval constraint, new real numbers are randomly generated and reassigned to the target dimension to generate the perturbation individual; The disturbed individual replaces the worst individual with the highest fitness value in the population for subsequent iterations, and the number of generations without improvement is reset to zero.

[0041] After each generation iteration, assuming the 42nd generation iteration ends, extract the minimum fitness value of the current population's 50 individuals. Record the best individual of the current generation .

[0042] The value of 1.34 is compared with the historical best fitness value, assuming the value recorded in generation 40 is also 1.34. Since they are equal, it indicates that the fitness value of the current best individual has not decreased, and the generation counter has not been improved. Increment by 1. When Reaching the preset stagnation algebra threshold When determining the target dimension, identify the dimension with the largest deviation between the predicted temperature and the set temperature. For example, if the absolute temperature difference of the 6th temperature zone is determined to be 2.6℃, which is higher than that of other temperature zones, then the 6th dimension is determined as the target dimension.

[0043] Within the closed-interval constraint range [-15.0, 15.0] kW, a new real number is randomly generated to reassign the target dimension. For example, if the original value is 8.5 kW, it is reassigned to -7.2 kW. The values ​​of the other 7 dimensions of the original individual are kept unchanged, and the value of the 6th dimension is replaced to generate a perturbed individual. This will disturb the individual. Replace the worst individual with the highest fitness value in the current population. For example, an individual with a fitness value of 9.5 will participate in subsequent iterations, and Reset to zero. (e.g.) Figure 2 As shown.

[0044] In some implementations, the step of decoding the optimal individual value into a heating power correction amount and sending it to the actuator when the termination condition is met includes: Determine whether the current iteration number has reached the preset maximum number of iterations, or whether the fitness value of the best individual in the current population is less than the allowable error threshold. If either condition is met, the termination condition is met and the iteration stops. Extract the optimal individual with the smallest fitness value from the current population, and then split and parse the real number sequence of each dimension contained in the individual according to the physical arrangement order of the temperature zone, and decode it into the heating power correction amount instruction corresponding to each temperature zone. The control network sends heating power correction commands for each temperature zone to the corresponding actuators in the field in parallel. The actuator adjusts the on / off duty cycle of the solid-state relay or the output power of the solid-state power regulator according to the received heating power correction command, thereby changing the equivalent heat generation of the corresponding temperature zone and completing the coordinated control of multiple temperature zones of the continuous dyeing machine.

[0045] Determine the current iteration number during the iteration process. Has the preset maximum number of iterations been reached? Set it to 200 or 300 times. If If the condition is met, the iteration stops. The optimal individual with the smallest fitness value is extracted from the current population; let the vector of this individual be denoted as . .

[0046] The vector contains a sequence of real numbers for each dimension, which is then sequentially split and parsed according to the physical arrangement of the temperature zones. This is decoded into heating power correction commands for each temperature zone, such as increasing the power by 3.5 kW in temperature zone 1 and decreasing it by 2.1 kW in temperature zone 2. These heating power correction commands for each temperature zone are then sent in parallel to the corresponding actuators in the field.

[0047] The output power of the solid-state power regulator is adjusted according to the received heating power correction command, or the duty cycle of the solid-state relay is adjusted to change the equivalent heating power. For example, the original setting is to provide 12kW, and it is adjusted to the equivalent heating power level of 15.5kW according to the command, thereby changing the heat generation of the corresponding temperature zone and completing the coordinated control of multiple temperature zones of the continuous dyeing machine.

[0048] In this application, a multi-point coordinated control system for continuous staining temperature zones is disclosed, the system comprising: The generation module is used to obtain the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, encode the heating power correction of each temperature zone into individuals according to the temperature zone order to construct the initial population, select individuals to be crossovered from the current population as target individuals, calculate the scaling factor based on the temperature zone deviation and the number of crossover adoptions in previous generations, and generate mutant individuals by combining the difference vector. The determination module is used to divide an individual into several continuous temperature zone segments based on the positive and negative changes in the difference of heating power correction between adjacent dimensions within the individual. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. The test individual is obtained by exchanging segments between the target individual and the mutated individual. The replacement module is used to input the individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones, the fitness value is calculated to select the best individual to retain. If the fitness value does not improve, a perturbation individual is generated to replace the worst individual in the population. When the iteration reaches the termination condition, the optimal individual is decoded into a heating power correction amount and sent to the actuator.

[0049] In some implementations, the step of obtaining the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous staining machine, and encoding the heating power correction amount of each temperature zone into individuals to construct the initial population in the order of the temperature zones includes: The target operating temperature of each temperature zone preset by the control system is obtained as the set temperature, and the real-time measurement value of the temperature sensor of each temperature zone is collected as the measured temperature. The absolute value of the difference between the measured temperature and the set temperature in each temperature zone is calculated as the temperature zone deviation; Determine the closed-range constraint range formed by the maximum allowable decrease threshold and the maximum increase threshold of the actuator in each temperature zone. The closed-range constraint range ensures that the sum of the current basic heating power of each temperature zone and any heating power correction amount is not less than zero and does not exceed the rated power limit of the corresponding actuator. Within the closed interval constraint range, a number of real numbers equal to the total number of temperature zones are randomly generated. The generated real numbers are arranged into row vectors according to the physical arrangement order of the temperature zones to construct an individual. Each dimension of the individual corresponds to the heating power correction amount of the corresponding temperature zone. The individuals are generated repeatedly until the total number of individuals reaches the preset population size, thus constructing the initial population.

[0050] In some implementations, the step of calculating the scaling factor based on the temperature zone deviation and the number of cross-adoptions in each generation, and then combining it with the difference vector to generate variant individuals, includes: Obtain the fitness values ​​of each individual in the current population and sort them in ascending order. Select the individual whose sorting position is half the population size as the basis vector. For any dimension, when the sum of all temperature zone deviations is greater than a preset minimum positive number, the ratio of the temperature zone deviation corresponding to that dimension to the sum of the temperature zone deviations is used as the basic scaling factor. When the sum of all temperature zone deviations is not greater than the preset minimum positive number, the basic scaling factor of each dimension is set to the preset equal value; Before the initial iteration, the historical adoption count of each dimension is initialized to zero, and the historical adoption count of that dimension that was adopted and retained during the historical iteration process is read as the cross-adoption count of each generation. The bias gain is determined according to the ratio of the preset constant to the sum of the historical adoption count and the preset ratio constant. The scaling factor of that dimension is obtained by multiplying the basic scaling factor by the bias gain. Randomly select two individuals from the current population that are different from the basis vector, and calculate the difference between the two in the same dimension as the difference vector; The difference vector is multiplied by the scaling factor of the corresponding dimension and then superimposed onto the basis vector to obtain the mutation value of each dimension. The mutation values ​​of each dimension are combined to generate mutated individuals. The mutation values ​​that exceed the closed interval constraint range are truncated at the boundary.

[0051] In some implementations, the individual is divided into several continuous temperature zone segments based on the positive and negative changes in the heating power correction difference between adjacent dimensions. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. Experimental individuals are obtained by exchanging segments between the target individual and the mutated individual. This includes: Calculate the difference in heating power correction between adjacent dimensions in the target individual. When two adjacent differences are both non-zero and the sign is reversed, set a segment point at the intersection of the corresponding dimensions. When the difference is zero, it is not used as a basis for reversing the sign. The target individual is divided into several consecutive temperature zones based on all the aforementioned segmentation points; Calculate the average of all temperature zone deviations within each temperature zone segment as the segment deviation mean, and select the largest segment deviation mean as the maximum segment deviation mean. When the mean of the maximum segment deviation is greater than the preset minimum positive number, the mean of the segment deviation is divided by the mean of the maximum segment deviation to obtain the segment crossover probability corresponding to each temperature zone segment; when the mean of the maximum segment deviation is not greater than the preset minimum positive number, the segment crossover probability corresponding to each temperature zone segment is set to a preset equal probability or a preset low probability. For each temperature zone segment, a random number between 0 and 1 is generated. When the random number is less than the corresponding segment crossover probability, the value of the corresponding temperature zone segment in the mutated individual is replaced with the corresponding position in the target individual. Otherwise, the original value of the temperature zone segment in the target individual is retained. After completing the crossover operation of all temperature zone segments, the experimental individual is obtained.

[0052] In some implementations, the step of inputting the individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone, and calculating the fitness value based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones to select the best individuals, includes: The heating power correction for each temperature zone corresponding to an individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone are all input into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. The difference between the set temperatures of two adjacent temperature zones is determined as the target temperature difference for the corresponding adjacent temperature zones. The sum of squares of the differences between the predicted temperature and the set temperature in each temperature zone of an individual is calculated and multiplied by a first weighting coefficient to obtain the first evaluation item; The second evaluation term is obtained by calculating the sum of squared deviations between the predicted temperature difference between adjacent temperature zones and the target temperature difference between the corresponding adjacent temperature zones, and multiplying the sum of squared deviations by the second weighting coefficient. The fitness value of an individual is obtained by adding the first evaluation item to the second evaluation item. The fitness values ​​of the experimental individuals and the target individuals are calculated respectively. The individuals with lower fitness values ​​are compared and retained into the next generation of the population to achieve the selective retention of individuals. If the fitness value of the experimental individual is less than the fitness value of the target individual, then the temperature range segments that have been exchanged in the experimental individual are checked. For temperature range segments that are taken from the mutant individuals and win the selection and are retained, the historical adoption count of the corresponding dimension in the statistical pool is incremented by 1, which is used to correct the scaling factor of the corresponding dimension in the next generation.

[0053] In some implementations, generating a perturbed individual to replace the worst individual in the population when the fitness value does not improve includes: After each generation iteration, record the best individual with the smallest fitness value in the current population and its fitness value; Compare the fitness value of the current generation's best individual with the historical best fitness value. If there is no decrease, increment the number of generations without improvement by 1. When the unimproved algebra reaches a preset stagnation algebra threshold, the target dimension with the largest deviation between the predicted temperature and the set temperature in the current best individual is determined. Within the closed interval constraint, new real numbers are randomly generated and reassigned to the target dimension to generate the perturbation individual; The disturbed individual replaces the worst individual with the highest fitness value in the population for subsequent iterations, and the number of generations without improvement is reset to zero.

[0054] In some implementations, the step of decoding the optimal individual value into a heating power correction amount and sending it to the actuator when the termination condition is met includes: Determine whether the current iteration number has reached the preset maximum number of iterations, or whether the fitness value of the best individual in the current population is less than the allowable error threshold. If either condition is met, the termination condition is met and the iteration stops. Extract the optimal individual with the smallest fitness value from the current population, and then split and parse the real number sequence of each dimension contained in the individual according to the physical arrangement order of the temperature zone, and decode it into the heating power correction amount instruction corresponding to each temperature zone. The control network sends heating power correction commands for each temperature zone to the corresponding actuators in the field in parallel. The actuator adjusts the on / off duty cycle of the solid-state relay or the output power of the solid-state power regulator according to the received heating power correction command, thereby changing the equivalent heat generation of the corresponding temperature zone and completing the coordinated control of multiple temperature zones of the continuous dyeing machine.

[0055] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for multi-point coordinated control of continuous staining temperature zone, characterized in that, include: Based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, the temperature zone deviation is obtained. The heating power correction of each temperature zone is encoded into individuals according to the temperature zone order to construct an initial population. Individuals to be crossovered are selected from the current population as target individuals. The scaling factor is calculated based on the temperature zone deviation and the number of crossover adoptions in previous generations. Combined with the difference vector, mutant individuals are generated. Individuals are divided into several continuous temperature zone segments based on the positive and negative changes in the difference of heating power correction between adjacent dimensions. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. Experimental individuals are obtained by exchanging segments between the target individual and the mutated individual. The individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone are input into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones, the fitness value is calculated to select the best individual to retain. If the fitness value does not improve, a perturbation individual is generated to replace the worst individual in the population. When the iteration reaches the termination condition, the optimal individual is decoded into a heating power correction amount and sent to the actuator.

2. The method according to claim 1, characterized in that, The method involves obtaining the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, and encoding the heating power correction amount of each temperature zone into individuals in temperature zone order to construct an initial population, including: The target operating temperature of each temperature zone preset by the control system is obtained as the set temperature, and the real-time measurement value of the temperature sensor of each temperature zone is collected as the measured temperature. The absolute value of the difference between the measured temperature and the set temperature in each temperature zone is calculated as the temperature zone deviation; Determine the closed-range constraint range formed by the maximum allowable decrease threshold and the maximum increase threshold of the actuator in each temperature zone. The closed-range constraint range ensures that the sum of the current basic heating power of each temperature zone and any heating power correction amount is not less than zero and does not exceed the rated power limit of the corresponding actuator. Within the closed interval constraint range, a number of real numbers equal to the total number of temperature zones are randomly generated. The generated real numbers are arranged into row vectors according to the physical arrangement order of the temperature zones to construct an individual. Each dimension of the individual corresponds to the heating power correction amount of the corresponding temperature zone. The individuals are generated repeatedly until the total number of individuals reaches the preset population size, thus constructing the initial population.

3. The method according to claim 1, characterized in that, The process of calculating the scaling factor based on temperature zone deviation and the number of cross-adoptions in previous generations, and generating variant individuals by combining the difference vector, includes: Obtain the fitness values ​​of each individual in the current population and sort them in ascending order. Select the individual whose sorting position is half the population size as the basis vector. For any dimension, when the sum of all temperature zone deviations is greater than a preset minimum positive number, the ratio of the temperature zone deviation corresponding to that dimension to the sum of the temperature zone deviations is used as the basic scaling factor. When the sum of all temperature zone deviations is not greater than the preset minimum positive number, the basic scaling factor of each dimension is set to the preset equal value; Before the initial iteration, the historical adoption count of each dimension is initialized to zero, and the historical adoption count of that dimension that was adopted and retained during the historical iteration process is read as the cross-adoption count of each generation. The bias gain is determined according to the ratio of the preset constant to the sum of the historical adoption count and the preset ratio constant. The scaling factor of that dimension is obtained by multiplying the basic scaling factor by the bias gain. Randomly select two individuals from the current population that are different from the basis vector, and calculate the difference between the two in the same dimension as the difference vector; The difference vector is multiplied by the scaling factor of the corresponding dimension and then superimposed onto the basis vector to obtain the mutation value of each dimension. The mutation values ​​of each dimension are combined to generate mutated individuals. The mutation values ​​that exceed the closed interval constraint range are truncated at the boundary.

4. The method according to claim 1, characterized in that, The individual is divided into several continuous temperature zone segments based on the positive and negative changes in the heating power correction difference between adjacent dimensions. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. Experimental individuals are obtained by exchanging segments between the target individual and the mutated individual. This includes: Calculate the difference in heating power correction between adjacent dimensions in the target individual. When two adjacent differences are both non-zero and the sign is reversed, set a segment point at the intersection of the corresponding dimensions. When the difference is zero, it is not used as a basis for reversing the sign. The target individual is divided into several consecutive temperature zones based on all the aforementioned segmentation points; Calculate the average of all temperature zone deviations within each temperature zone segment as the segment deviation mean, and select the largest segment deviation mean as the maximum segment deviation mean. When the mean of the maximum segment deviation is greater than the preset minimum positive number, the mean of the segment deviation is divided by the mean of the maximum segment deviation to obtain the segment crossover probability corresponding to each temperature zone segment; when the mean of the maximum segment deviation is not greater than the preset minimum positive number, the segment crossover probability corresponding to each temperature zone segment is set to a preset equal probability or a preset low probability. For each temperature zone segment, a random number between 0 and 1 is generated. When the random number is less than the corresponding segment crossover probability, the value of the corresponding temperature zone segment in the mutated individual is replaced with the corresponding position in the target individual. Otherwise, the original value of the temperature zone segment in the target individual is retained. After completing the crossover operation of all temperature zone segments, the experimental individual is obtained.

5. The method according to claim 1, characterized in that, The process involves inputting the individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones, a fitness value is calculated to select the best individuals for retention. This includes: The heating power correction for each temperature zone corresponding to an individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone are all input into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. The difference between the set temperatures of two adjacent temperature zones is determined as the target temperature difference for the corresponding adjacent temperature zones. The sum of squares of the differences between the predicted temperature and the set temperature in each temperature zone of an individual is calculated and multiplied by a first weighting coefficient to obtain the first evaluation item; The second evaluation term is obtained by calculating the sum of squared deviations between the predicted temperature difference between adjacent temperature zones and the target temperature difference between the corresponding adjacent temperature zones, and multiplying the sum of squared deviations by the second weighting coefficient. The fitness value of an individual is obtained by adding the first evaluation item to the second evaluation item. The fitness values ​​of the experimental individuals and the target individuals are calculated respectively. The individuals with lower fitness values ​​are compared and retained into the next generation of the population to achieve the selective retention of individuals. If the fitness value of the experimental individual is less than the fitness value of the target individual, then the temperature range segments that have been exchanged in the experimental individual are checked. For temperature range segments that are taken from the mutant individuals and win the selection and are retained, the historical adoption count of the corresponding dimension in the statistical pool is incremented by 1, which is used to correct the scaling factor of the corresponding dimension in the next generation.

6. The method according to claim 1, characterized in that, The step of generating a perturbation individual to replace the worst individual in the population when the fitness value does not improve includes: After each generation iteration, record the best individual with the smallest fitness value in the current population and its fitness value; Compare the fitness value of the current generation's best individual with the historical best fitness value. If there is no decrease, increment the number of generations without improvement by 1. When the unimproved algebra reaches a preset stagnation algebra threshold, the target dimension with the largest deviation between the predicted temperature and the set temperature in the current best individual is determined. Within the closed interval constraint, new real numbers are randomly generated and reassigned to the target dimension to generate the perturbation individual; The disturbed individual replaces the worst individual with the highest fitness value in the population for subsequent iterations, and the number of generations without improvement is reset to zero.

7. The method according to claim 1, characterized in that, The step of decoding the optimal individual value into a heating power correction amount and sending it to the actuator when the termination condition is met during iteration includes: Determine whether the current iteration number has reached the preset maximum number of iterations, or whether the fitness value of the best individual in the current population is less than the allowable error threshold. If either condition is met, the termination condition is met and the iteration stops. Extract the optimal individual with the smallest fitness value from the current population, and then split and parse the real number sequence of each dimension contained in the individual according to the physical arrangement order of the temperature zone, and decode it into the heating power correction amount instruction corresponding to each temperature zone. The control network sends heating power correction commands for each temperature zone to the corresponding actuators in the field in parallel. The actuator adjusts the on / off duty cycle of the solid-state relay or the output power of the solid-state power regulator according to the received heating power correction command, thereby changing the equivalent heat generation of the corresponding temperature zone and completing the coordinated control of multiple temperature zones of the continuous dyeing machine.

8. A multi-point coordinated control system for continuous staining temperature zones, characterized in that, include: The generation module is used to obtain the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, encode the heating power correction of each temperature zone into individuals according to the temperature zone order to construct the initial population, select individuals to be crossovered from the current population as target individuals, calculate the scaling factor based on the temperature zone deviation and the number of crossover adoptions in previous generations, and generate mutant individuals by combining the difference vector. The determination module is used to divide an individual into several continuous temperature zone segments based on the positive and negative changes in the difference of heating power correction between adjacent dimensions within the individual. The segment crossover probability is determined based on the ratio of the average temperature zone deviation within each segment to the average deviation of the largest segment. The test individual is obtained by exchanging segments between the target individual and the mutated individual. The replacement module is used to input the individual, the measured temperature of each temperature zone, and the current basic heating power of each temperature zone into the continuous dyeing machine temperature prediction model to obtain the predicted temperature of each temperature zone. Based on the predicted temperature, the set temperature, and the target temperature difference between adjacent temperature zones, the fitness value is calculated to select the best individual to retain. If the fitness value does not improve, a perturbation individual is generated to replace the worst individual in the population. When the iteration reaches the termination condition, the optimal individual is decoded into a heating power correction amount and sent to the actuator.

9. The system according to claim 8, characterized in that, The method involves obtaining the temperature zone deviation based on the measured temperature and set temperature of each temperature zone of the continuous dyeing machine, and encoding the heating power correction amount of each temperature zone into individuals in temperature zone order to construct an initial population, including: The target operating temperature of each temperature zone preset by the control system is obtained as the set temperature, and the real-time measurement value of the temperature sensor of each temperature zone is collected as the measured temperature. The absolute value of the difference between the measured temperature and the set temperature in each temperature zone is calculated as the temperature zone deviation; Determine the closed-range constraint range formed by the maximum allowable decrease threshold and the maximum increase threshold of the actuator in each temperature zone. The closed-range constraint range ensures that the sum of the current basic heating power of each temperature zone and any heating power correction amount is not less than zero and does not exceed the rated power limit of the corresponding actuator. Within the closed interval constraint range, a number of real numbers equal to the total number of temperature zones are randomly generated. The generated real numbers are arranged into row vectors according to the physical arrangement order of the temperature zones to construct an individual. Each dimension of the individual corresponds to the heating power correction amount of the corresponding temperature zone. The individuals are generated repeatedly until the total number of individuals reaches the preset population size, thus constructing the initial population.

10. The system according to claim 8, characterized in that, The process of calculating the scaling factor based on temperature zone deviation and the number of cross-adoptions in previous generations, and generating variant individuals by combining the difference vector, includes: Obtain the fitness values ​​of each individual in the current population and sort them in ascending order. Select the individual whose sorting position is half the population size as the basis vector. For any dimension, when the sum of all temperature zone deviations is greater than a preset minimum positive number, the ratio of the temperature zone deviation corresponding to that dimension to the sum of the temperature zone deviations is used as the basic scaling factor. When the sum of all temperature zone deviations is not greater than the preset minimum positive number, the basic scaling factor of each dimension is set to the preset equal value; Before the initial iteration, the historical adoption count of each dimension is initialized to zero, and the historical adoption count of that dimension that was adopted and retained during the historical iteration process is read as the cross-adoption count of each generation. The bias gain is determined according to the ratio of the preset constant to the sum of the historical adoption count and the preset ratio constant. The scaling factor of that dimension is obtained by multiplying the basic scaling factor by the bias gain. Randomly select two individuals from the current population that are different from the basis vector, and calculate the difference between the two in the same dimension as the difference vector; The difference vector is multiplied by the scaling factor of the corresponding dimension and then superimposed onto the basis vector to obtain the mutation value of each dimension. The mutation values ​​of each dimension are combined to generate mutated individuals. The mutation values ​​that exceed the closed interval constraint range are truncated at the boundary.

Images