Task flow-based air compression system control method

Abstract

The application relates to a kind of air compression system control methods based on task flow, comprising: modeling the task flow mode of gas end, according to air compression system task flow model, improved Harris hawk algorithm is used to optimize air compression system control strategy;Build air compression system model;Initialize the global variable of improved Harris hawk algorithm;The population initialization of original Harris hawk is optimized through reverse mapping;Iterative optimization is carried out to the initialized Harris hawk population;Individuals in the optimized population are executed under the partition elitism strategy of variable isolation;The optimization variable of the best individual is used to control the air tank output pressure of air compression system, the start-stop time and output pressure of each air compressor.Compared with the prior art, the application can efficiently optimize the task flow under the consideration of start-stop loss, and can also optimize the temporary task in real time, compared with the previous air compression system control optimization, more detailed and more flexible.

Description

Technical Field

[0001] This invention relates to the field of compressed air system design and optimization, and in particular to a task flow-based compressed air system control method. Background Technology

[0002] Compressed air systems are an indispensable part of industrial production and a major energy consumer in factories. Optimizing the control methods of compressed air systems can significantly reduce their total energy consumption during production. However, the optimization of compressed air system control typically relies on on- and off-peak operation based on a given load table. This approach cannot provide more precise control over the air compressor, such as adjusting its output pressure, and it also cannot adapt to the current workflow patterns, hindering the large-scale application of intelligent scheduling software in industrial production.

[0003] Therefore, the following technical directions still need improvement in the control optimization of air compressor systems:

[0004] Precise control of air compressor output pressure: Current control methods cannot precisely adjust the output pressure of air compressors, and more refined control methods need to be researched.

[0005] To adapt to the current task flow model: the control optimization of the air compressor system needs to adapt to the large-scale application of intelligent scheduling software in current industrial production, and it is necessary to study the collaborative control method with intelligent scheduling software to achieve more efficient production.

[0006] Reduce total energy consumption: Although optimizing the control of the compressed air system can reduce total energy consumption, further research is needed on control methods to achieve a greater degree of energy reduction. Summary of the Invention

[0007] The purpose of this invention is to overcome at least one of the defects of the prior art and provide a task flow-based air compressor system control method. This method can perform efficient global optimization of the task flow while taking into account start-up and shutdown losses, and can also perform real-time optimization for temporary tasks. Compared with previous air compressor system control optimization, this method is more detailed and flexible.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] This invention provides a task-flow-based air compressor system control method, comprising the following steps:

[0010] S1: Model the gas-consuming end task flow mode, and optimize the air compressor system control strategy based on the air compressor system task flow model using the improved Harris Eagle algorithm;

[0011] S2: Construct an air compressor system model, which consists of three parts: an air compressor, an air tank, and a pipeline network;

[0012] S3: Initialize global variables for the improved Harris Eagle algorithm;

[0013] S4: Optimize the initialization of the original Harris Hawk population through reverse mapping;

[0014] S5: Iteratively optimize the initialized Harris Eagle population;

[0015] S6: Implement a partitioned elitist strategy with variable isolation for individuals in the optimized population;

[0016] S7: Complete the iteration process and use the optimization variables of the best historical individual to control the output pressure of the air storage tank of the air compressor system, the start-up and shutdown time and output pressure of each air compressor.

[0017] Furthermore, in S1, the modeling process for the gas-consuming task flow pattern specifically includes:

[0018] Suppose at a certain moment, a gas user submits a gas consumption task as follows:

[0019]

[0020] Among them, user i This represents the i-th gas-consuming terminal. Indicates user i The gas demand at time j, where h represents the maximum time of the task flow at this moment;

[0021] For the period from 1 hour to h, there are n gas-consuming terminals with the following task flow model:

[0022]

[0023] Furthermore, in S2, the process of constructing the compressed air system model includes:

[0024] Construct the pipeline network according to the topology of the compressed air system;

[0025] The output pressure of the gas-consuming pipeline is obtained by calculating the pressure drop of a single pipe.

[0026] The single-tube voltage drop is calculated using the overall method, as shown in the following formula:

[0027]

[0028] in:

[0029]

[0030]

[0031]

[0032]

[0033] Where: Q_m is the mass flow rate of the single-tube structure parameter, D is the inner diameter of the single-tube structure parameter, ρ is the density at the inlet end of the single-tube structure parameter, v is the gas velocity, and nodein mu Here, denoted as the inlet dynamic viscosity of the single-tube structure parameter, Re as the Reynolds number, 1e6 as scientific notation (10 to the power of 6), zeta as the local drag coefficient of the single-tube structure parameter, k as the roughness inside the single-tube structure parameter, ΔH as the absolute height difference of the single-tube structure parameter, and g as the gravitational constant (taken as 9.8).

[0034] The energy consumption of the air compressor is as follows:

[0035]

[0036] Where W is the air compressor energy consumption, γ is the adiabatic coefficient (taken as 1.4 for air), R is the gas constant (taken as 287 for air), and T... in For the inlet temperature, P in For the inlet pressure, take 0.1 kPa, η s For isentropic efficiency, we take 0.8, P out Due to export pressure, Air mass flow rate.

[0037] Furthermore, in S3, the global variables of the Harris Eagle algorithm include the number of Harris Eagles search_num, the dimension dim, the upper limit and lower limit of the variable interval UB and max_iter;

[0038] Each Harris Eagle corresponds to an air compressor system control strategy, which includes the air tank output pressure and the air compressor start-up and shutdown time.

[0039] During the initialization of global variables for the improved Harris Eagle algorithm, the air compressor output pressure is used as the optimization variable:

[0040]

[0041] in, To optimize the variable vector, tank n This refers to gas storage tank number n. c represents the output pressure of gas storage tank n. n This indicates air compressor number n. This indicates the start-up time of air compressor number n. This indicates the shutdown time of air compressor number n. This indicates the output pressure of air compressor number n;

[0042] in, As a time variable, it is represented using the following decimal method:

[0043]

[0044] Where hh = 0, ..., 24 represents hours, and mm = 0, ..., 59 represents minutes.

[0045] Furthermore, in S4, the process of optimizing the original Harris Hawk population initialization through reverse mapping includes:

[0046] S4-1, the optimized population is randomly generated based on the number of Harris eagles search_num, dimension dim, upper limit of variable interval UB and lower limit LB, as shown in formula (4):

[0047] pop r =random(search_num,dim)*(UB-LB)+LB(4)

[0048] Where pop_r is a randomly generated optimization population, random() generates a random decimal number in the interval 0 to 1 that follows a uniform distribution, and random(search_num,dim) generates a random decimal matrix with the number of rows search_num and the number of columns dim;

[0049] S4-2, Calculate pop r The fitness value of each individual, i.e. the total energy consumption generated by each air compressor system control strategy, is shown in formula (5):

[0050]

[0051] Where fitness() is the function for calculating the total energy consumption of the compressed air system, i.e., the fitness calculation function. For pop r The i-th control policy, fit r For random population fitness, Let be the total energy consumption of the i-th control strategy, which is the fitness value of the i-th control strategy;

[0052] The reverse population is generated according to formula (6), and the fitness value of the reverse population is calculated according to formula (7). Formula (6) is shown below:

[0053]

[0054] Among them, pop o For the generated reverse population;

[0055] Formula (7) is:

[0056]

[0057] Among them, fit o For reverse population fitness, Let be the fitness value of the i-th reverse population;

[0058] S4-3, fit the random population fitness r and reverse population fitness o Merge into fit r～o Pop a random population r and reverse population pop o merge into pop r～o r~o indicates that the reversed population is vertically concatenated after the random population, for fit r～o Sort the control strategies and select the search_num strategies with the lowest total energy consumption to form the initial population pop that needs to be iterated.

[0059] Furthermore, in S4, the iterative optimization process for the initialized Harris Eagle population includes:

[0060] Different evolutionary strategies can be selected based on the magnitude of the E value calculated using formula (8):

[0061]

[0062] Where cur_iter is the current iteration number, and E0 is a random number in the range (-1, 1).

[0063] Furthermore, in S4, during the selection of different evolutionary strategies:

[0064] When the absolute value of E is greater than or equal to 1, the individual chooses the evolutionary strategy shown in formula (9):

[0065]

[0066] in, For the evolution of offspring, As the best individual at present, For the current individual, Let be a randomly selected individual, e be the natural constant, and q, r1, r2, r3, r4 be random decimals within the range (0, 1). This represents the average position of the current group. This represents the absolute distance between the current individual and the optimal individual. and The calculation formula is as follows:

[0067]

[0068]

[0069] Furthermore, in S4, during the selection of different evolutionary strategies:

[0070] When E is less than 1, an evolutionary strategy is executed. This strategy includes four specific strategies, and the choice of which evolutionary strategy to execute is determined by the magnitude of a random decimal r_select within the range (0,1) and the value of E.

[0071] (1) If r_select is greater than or equal to 0.5 and the absolute value of E is greater than or equal to 0.5, then the evolutionary strategy shown in formula (10) is executed;

[0072]

[0073] Where r5 is a random decimal number within (0,1), The distance between the current best individual and the current individual;

[0074] (2) If r_select is greater than or equal to 0.5 and the absolute value of E is less than 0.5, then the evolutionary strategy shown in formula (11) is executed.

[0075]

[0076] (3) If r_select is less than 0.5 and the absolute value of E is greater than or equal to 0.5, then the evolutionary strategy shown in formula (12) is executed.

[0077]

[0078] Where r6 is a random decimal number within (0,1);

[0079] (4) If r_select is less than 0.5 and the absolute value of E is less than 0.5, then the evolutionary strategy shown in formula (13) is executed;

[0080]

[0081] Where r7 is a random decimal number within (0,1).

[0082] Furthermore, in S5, the partitioned elitist strategy process under variable isolation includes:

[0083] For the output pressure of the gas storage tank, a disturbance strategy based on Levi flight is adopted, specifically using the following formula:

[0084]

[0085] in, Given a dim-dimensional vector of random decimals between 0 and 1, LF(dim) is obtained by the following formula:

[0086]

[0087]

[0088] Where u and v are random decimals between 0 and 1, Γ is the gamma function, and β takes the value of 1.5;

[0089] When implementing an elitist strategy, the following formula is used:

[0090]

[0091] Furthermore, in S5, the partitioned elitist strategy process under variable isolation also includes:

[0092] The output pressure of the air compressor adopts a differential evolution strategy. The differential evolution strategy is used to determine the relationship between the air compressor's output pressure and its start / stop time, as shown in the following formula:

[0093]

[0094] Wherein, ind1, ind2, ind3, ind4, ind5, and ind6 are random integers between 0 and search_num;

[0095] When implementing an elitist strategy, the following formula is used:

[0096]

[0097] Compared with the prior art, the present invention has the following technical advantages:

[0098] A) The optimization variable ranges of the air compressor system differ. For example, the upper and lower limits of the air tank output pressure are typically between 600-900, while the start-up and shutdown times of the air compressor range from 0-24 and have only two decimal places. Therefore, the air compressor system control optimization variable space is a non-uniform and discontinuous variable space. This means that the randomly generated initial population relied upon by the original Harris Eagle algorithm is not suitable for the air compressor system control optimization problem. This is because the randomly generated initial population cannot be uniformly distributed in the non-uniform and discontinuous air compressor system control optimization variable space. The randomly generated initial population will cause the original Harris Eagle algorithm to be limited to local search during the optimization process, failing to achieve a true global search, resulting in premature convergence and failure to find the minimum energy consumption. Therefore, this invention introduces inverse mapping for population initialization. Through inverse mapping, prior information about the optimization variables can be used to uniformly map the population into the air compressor system optimization variable space, preventing the population from getting trapped in local optima and premature convergence, thus improving the optimization effect of this invention on the air compressor system control strategy.

[0099] B) In the iterative process of air compressor system control optimization based on task flow, the most computationally time-consuming part is fitness calculation, that is, the calculation of the total energy consumption of the air compressor system under this control strategy. In the original Harris Eagle algorithm, in technical solution S5 (3)(4), a Levee instant search strategy is set, that is, a new individual is generated instantly through the Levee flight strategy, and its fitness value is calculated and compared with the fitness value of the current individual. If the fitness value of the new individual is less than the fitness value of the current individual, the new individual generated by the Levee flight strategy is retained. However, this method is not suitable for air compressor system control optimization. The reasons are as follows:

[0100] (1) The Levy flight strategy is not applicable to the control optimization of air compressor systems. The Levy flight strategy essentially assumes that the optimization variables follow a uniform distribution, thereby generating a series of random decimals for local search. However, the space of optimization variables for air compressor system control is not uniformly distributed, therefore the Levy flight strategy is not applicable to the control optimization of air compressor systems.

[0101] (2) The original Harris Eagle algorithm requires calculating the fitness value 4 times in each iteration of the instant elitism, which is undoubtedly time-consuming and inefficient and does not conform to the actual application scenario of the air compressor system.

[0102] Therefore, this invention modifies the iterative structure based on the properties of the optimization variables, eliminates immediate elitism, and, when used in conjunction with other improvement methods, simplifies the process and enhances the effectiveness of the invention in practical applications.

[0103] C) Due to the varying properties and intervals of the variables in the air compressor system optimization variable space, it is non-uniform and discontinuous. Therefore, this invention proposes a differential elite search strategy under variable isolation to improve the update granularity of the evolutionary strategy, effectively addressing the weakness of the Levy flight strategy in the original Harris Eagle algorithm, which struggles to function effectively in a non-uniform and discontinuous optimization variable space, thus enhancing the local search capability of the improved Harris Eagle algorithm. For the gas tank output pressure variable, since its variable space is continuous and requires random perturbation for local search, this invention employs Levy flight to improve the local search capability of the gas tank pressure variable. For discrete variables such as start-up and shutdown times, this invention uses a differential local search strategy to increase the local search capability for the air compressor start-up and shutdown times.

[0104] D) The task flow model proposed in this invention facilitates the practical implementation of air compressor system control optimization methods. In actual engineering projects, the gas user typically submits a gas consumption task to the air compressor station, which then coordinates the operation of various equipment such as air compressors and air tanks. Previous inventions involving air compressor system control optimization did not consider the actual task flow, leading to difficulties in timely and effective handling of air compressor system optimization control problems in practical engineering projects. Therefore, this invention, combined with practical engineering, proposes a task flow-based air compressor system control optimization, establishing an air compressor system task flow model. This effectively addresses the pain point of difficulty in implementation, and the task flow model has strong scalability and can be integrated with automatic scheduling software. Attached Figure Description

[0105] Figure 1 This is a schematic diagram of the compressed air system structure in Application Example 1. Detailed Implementation

[0106] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms, and other features not explicitly described in this technical solution are considered common technical features disclosed in the prior art.

[0107] Example 1

[0108] The air compressor system control method based on task flow in this technical solution includes the following steps:

[0109] S1:

[0110] First, the gas-consuming task flow pattern is modeled as follows:

[0111] Suppose at a certain moment, a gas user submits a gas consumption task as follows:

[0112]

[0113] Among them, user i This represents the i-th gas-consuming terminal. Indicates user i The gas demand at time j, where h represents the maximum time of the task flow at that moment.

[0114] Based on the above formula, for the period from 1 hour to h, there are n task flow models for gas-consuming terminals as follows:

[0115]

[0116] Subsequently, based on the air compressor system task flow model shown in formula (1), the improved Harris Eagle algorithm is used to optimize the air compressor system control strategy in order to reduce the total energy consumption of the air compressor system.

[0117] S2:

[0118] Construct an air compressor system model, which consists of three parts: an air compressor, an air receiver, and a piping network. The air compressor compresses air into the air receiver, and the air receiver outputs the gas to the user end through the piping network.

[0119] The pipeline network is constructed based on the topology of the compressed air system. The output pressure of the gas-consuming pipeline is obtained by calculating the pressure drop of a single pipe. The pressure drop of a single pipe is calculated using the overall method, and the formula is as follows:

[0120]

[0121] in:

[0122]

[0123]

[0124]

[0125]

[0126] Q_m is the mass flow rate of a single-tube structure, D is the pipe inner diameter of a single-tube structure, ρ is the density at the inlet of a single-tube structure, v is the gas velocity, and nodein mu Here, denoted as the inlet dynamic viscosity of the single-tube structure parameter, Re as the Reynolds number, 1e6 as scientific notation (10 to the power of 6), zeta as the local drag coefficient of the single-tube structure parameter, k as the roughness inside the single-tube structure parameter, ΔH as the absolute height difference of the single-tube structure parameter, and g as the gravitational constant, taken as 9.8.

[0127] The air compressor's energy consumption is calculated using an isentropic compression model and isentropic efficiency. The isentropic compression model is shown below:

[0128]

[0129] Substituting the following equation into the isentropic compression model

[0130]

[0131] We can obtain:

[0132]

[0133] Where W is the air compressor energy consumption, γ is the adiabatic coefficient (taken as 1.4 for air), R is the gas constant (taken as 287 for air), and T... in For the inlet temperature, P in For the inlet pressure, take 0.1 kPa, which is atmospheric pressure, η s For isentropic efficiency, we take 0.8, P out This is due to export pressure.

[0134] The viscosity and density calculations required for the compressed air system are performed by calling the open-source property calculation library CoolProp.

[0135] S3:

[0136] Initialize global variables for the improved Harris Hawk algorithm, such as the number of Harris Hawks search_num, where each Harris Hawk represents a control strategy for the compressed air system. Also include the dimension dim, the upper limit and lower limit of the variable range UB and LB, and the maximum number of iterations max_iter.

[0137] As shown in S2, the air compressor system needs to control the air tank output pressure, the air compressor start-up and shutdown time, and the output pressure. Therefore, an air compressor system control strategy includes the air tank output pressure and the air compressor start-up and shutdown time. During the initialization of the global variables in the improved Harris Eagle algorithm, the air compressor output pressure is used as the optimization variable, as shown below:

[0138]

[0139] in, To optimize the variable vector, tank n This refers to gas storage tank number n. c represents the output pressure of gas storage tank n. n This indicates air compressor number n. This indicates the start-up time of air compressor number n. This indicates the shutdown time of air compressor number n. This indicates the output pressure of air compressor number n.

[0140] As a time variable, it is represented using the following decimal method:

[0141]

[0142] Where hh = 0, ..., 24 represents hours, and mm = 0, ..., 59 represents minutes. This reduces the range of the time variable, speeding up the optimization process and improving the results.

[0143] S4:

[0144] This invention optimizes the initialization of the original Harris Eagle population through reverse mapping, and the specific process is as follows:

[0145] First, an optimized population is randomly generated based on the number of Harris eagles (search_num), the dimension (dim), the upper limit of the variable interval (UB), and the lower limit (LB), as shown in formula (4):

[0146] pop r=random(search_num,dim)*(UB-LB)+LB(4)

[0147] Where pop_r is a randomly generated optimization population, random() generates a random decimal number in the interval 0 to 1 that follows a uniform distribution, and random(search_num,dim) generates a random decimal matrix with the number of rows search_num and the number of columns dim.

[0148] Next, calculate pop. r The fitness value of each individual, i.e. the total energy consumption generated by each air compressor system control strategy, is shown in formula (5):

[0149]

[0150] Where fitness() is the function for calculating the total energy consumption of the compressed air system, i.e., the fitness calculation function. For pop r The i-th control policy, fit r For random population fitness, Let be the total energy consumption of the i-th control strategy, which is the fitness value of the i-th control strategy.

[0151] The reverse population is generated according to formula (6), and the fitness value of the reverse population is calculated according to formula (7). Formula (6) is shown below:

[0152]

[0153] Among them, pop o This is the generated reverse population. Formula (7) is shown below:

[0154]

[0155] Among them, fit o For reverse population fitness, Let be the fitness value of the i-th reverse population.

[0156] Finally, the fitness of the random population is fitted. r and reverse population fitness o Merge into fit r～o Pop a random population r and reverse population pop o merge into pop r～o , r ~ o indicates that the reversed population is vertically concatenated after the random population. For fit r～oSort the control strategies and select the search_num strategies with the lowest total energy consumption to form the initial population pop that needs to be iterated.

[0157] By using the reverse mapping technique, the original search space can be redefined, and the initial position of the Harris Eagle can be mapped from the original search space to the new search space. Then, the position of the Harris Eagle can be mapped back from the new search space to the original search space. This optimizes the initialization of the population in the Harris Eagle algorithm, accelerates the convergence of the search to the global optimum, and improves the search accuracy.

[0158] S5:

[0159] The initial population pop is iteratively optimized. Different evolutionary strategies are selected based on the value of E obtained by formula (8).

[0160]

[0161] Where, cur_iter is the current iteration number, and E0 is a random number in the range (-1, 1). When the absolute value of E is greater than or equal to 1, the individual chooses the evolutionary strategy shown in formula (9):

[0162]

[0163] in, For the evolution of offspring, The optimal individual (i.e., the control strategy that generates the lowest total energy consumption) is selected. For the current individual, Let be a randomly selected individual, e be the natural constant, and q, r1, r2, r3, r4 be random decimals within the range (0, 1). This represents the average position of the current group. This represents the absolute distance between the current individual and the optimal individual. and The calculation formula is as follows:

[0164]

[0165]

[0166] When E is less than 1, the development strategy is executed. There are four development strategies, and the choice of which evolutionary strategy to execute is determined by the size of a random decimal r_select in (0,1) and the value of E.

[0167] (1) r_select is greater than or equal to 0.5, and the absolute value of E is greater than or equal to 0.5. At this time, the evolutionary strategy shown in formula (10) is executed.

[0168]

[0169] Where r5 is a random decimal number within (0,1), This represents the distance between the current best individual and the current individual.

[0170] (2) r_select is greater than or equal to 0.5, and the absolute value of E is less than 0.5. At this time, the evolutionary strategy shown in formula (11) is executed.

[0171]

[0172] (3) r_select is less than 0.5, and the absolute value of E is greater than or equal to 0.5. At this time, the evolutionary strategy shown in formula (12) is executed.

[0173]

[0174] Where r6 is a random decimal number within (0,1).

[0175] (4) r_select is less than 0.5, and the absolute value of E is less than 0.5. At this time, the evolutionary strategy shown in formula (13) is executed.

[0176]

[0177] Where r7 is a random decimal number within (0,1).

[0178] In this invention, since the calculation of the total energy consumption of the air compressor system takes too long, and the one-dimensional decimal generated by the Levi flight strategy is not suitable for the air compressor start-up and shutdown time in the optimization variables, the evolution strategies of (3) and (4) of the original Harris Eagle algorithm are improved, and the Levi flight strategy and the instant elite strategy in the original Harris Eagle are cancelled.

[0179] S6:

[0180] After an individual completes S5, the partitioned elitism strategy under variable isolation proposed in this invention is implemented. Variable isolation means designing an evolutionary strategy that suits the properties of each optimization variable.

[0181] For the output pressure of the gas storage tank, this invention adopts a disturbance strategy based on Levi flight, as shown in the following formula:

[0182]

[0183] in, Given a dim-dimensional vector of random decimals between 0 and 1, LF(dim) is obtained by the following formula:

[0184]

[0185]

[0186] Where u and v are random decimals between 0 and 1, Γ is the gamma function, and β takes the value 1.5. Substitution Make him The formula for applying an elitist strategy to them is as follows:

[0187]

[0188] Considering the pulsating phenomenon of the air compressor's output pressure, a differential evolution strategy is adopted for the air compressor's output pressure, rather than the LEVEL strategy. For the air compressor's output pressure and start-up / stop time, this invention uses a differential evolution strategy, as shown in the following formula:

[0189]

[0190] Where ind1, ind2, ind3, ind4, ind5, and ind6 are random integers between 0 and search_num. Substitution Make him The formula for applying an elitist strategy to them is as follows:

[0191]

[0192] S7:

[0193] After the iteration is completed, the optimization variables of the best historical individual are used to control the output pressure of the air storage tank of the air compressor system, the start-up and shutdown time of each air compressor, and the output pressure.

[0194] Application Example 1

[0195] This application example provides a specific implementation case of an intelligent control method for an air compressor system based on task flow. A schematic diagram of the air compressor system in the case is shown below. Figure 1 As shown.

[0196] S1:

[0197] Based on technical solution S1, the task flow model is modeled.

[0198] Depend on Figure 1 The task flow model for the four gas-consuming terminals within three hours is as follows:

[0199]

[0200]

[0201]

[0202]

[0203] Abstracted into the following matrix:

[0204]

[0205] S2:

[0206] Based on technical solution S2, the air compressor system is modeled. This case is an example. Figure 1 As shown, an air compressor is responsible for supplying air to the air storage tank and then to four air-consuming terminals.

[0207] The air compressor is modeled according to formula (3), and its total energy consumption is calculated. The pipeline network is modeled, with the flow direction of each pipe known. There are 7 pipes in the network, and the specific parameters are shown in the table below:

[0208] Table 1 Specific Data of Pipeline Network

[0209]

[0210] Based on the specific data of the pipeline network in Table 1 and formula (2), calculate the real-time pressure drop of each single pipe in the pipeline network according to the input pressure.

[0211] S3:

[0212] The case involves one air compressor and one air receiver, therefore the optimization variables are as follows:

[0213]

[0214] The flow demand is known and fixed. In this case, it is assumed that the pipeline mass flow rate is conserved, the pressure at the tee branch is equal, and the constraint is to meet the minimum gas demand at each gas consumption end in the task flow.

[0215] Initialize the parameters of the improved Harris Hawk optimization algorithm, and search for the number of Harris Hawks. num =30, each Harris Eagle is a control strategy for an air compressor system, dimension dim=4, upper limit of variable interval UB=[950,4,4,5000] and lower limit LB=[600,1,1,1000], maximum number of iterations max_iter=1000.

[0216] S4:

[0217] According to technical solutions S4-S7, after the improvement of the Harris Eagle algorithm is completed, the best historical individual will be used as the final air compressor system control strategy.

[0218] The optimization results are shown in Table 2:

[0219] Gas tank output pressure: 723.89 kPa Air compressor output pressure: 3244.63 kPa Opening time: 1.00 Shutdown time: 3.89 Total energy consumption of air compressor 8972.8kW

[0220] The results are compared with those of the TPE algorithm and the original Harris Eagle algorithm, as shown in Table 3.

[0221]

[0222] This implementation case demonstrates that the improved Harris Eagle algorithm in this invention is significantly better than the original Harris Eagle algorithm and the TPE algorithm, proving that the improved Harris Eagle algorithm proposed in this invention has certain advantages and advanced features.

[0223] The above description of the embodiments is provided to enable those skilled in the art to understand and use the invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the invention should be within the protection scope of the present invention.

Claims

1. A method for controlling an air compression system based on a task flow, characterized in that, The method comprises the following steps: S1: modeling the gas end task flow mode, and optimizing the air compression system control strategy according to the air compression system task flow model by using the improved Harris hawk algorithm; S2: constructing an air compression system model, wherein the air compression system comprises an air compressor, an air tank and a pipe network; S3: initializing global variables of the improved Harris hawk algorithm; S4: optimizing population initialization of the original Harris hawk through reverse mapping; S5: iteratively optimizing the initialized Harris hawk population; S6: executing a partition elitism strategy under variable isolation on individuals in the optimized population; S7: completing the iteration process, and using optimized variables of the historical best individual to control the air tank output pressure, the start and stop time and the output pressure of each air compressor of the air compression system; In S1, the process of modeling the gas end task flow mode specifically comprises: It is assumed that, at a certain moment, a gas end submits a gas task as follows: wherein, denotes the gas end, denotes the gas demand at the time t, denotes the maximum time of the task flow at this time; From 1 to h, the task flow model of n gas ends is as follows: 。 2. The method according to claim 1, wherein, In S2, the process of constructing an air compression system model comprises: Building a pipe network according to the air compression system pipe network topology structure; Obtaining the gas end pipe output pressure by calculating the single pipe pressure drop; The single pipe pressure drop is calculated by the whole method, and the formula is as follows: Wherein: wherein: is a single pipe structure parameter mass flow rate, is a single pipe structure parameter pipe inner diameter, is a single pipe structure parameter inlet density, is a gas velocity, is a single pipe structure parameter inlet dynamic viscosity, is a Reynolds number, is a scientific notation, representing 10 to the power of 6, is a single pipe structure parameter local resistance coefficient, is a single pipe structure parameter pipe inner roughness, is a single pipe structure parameter absolute height difference, is a gravitational constant, taken as 9.8; Wherein, the air compressor energy consumption is: wherein, is the power consumption of the air compressor, is the adiabatic coefficient, taken as 1.4 for air, is the gas constant, taken as 287 for air, is the inlet temperature, is the inlet pressure, taken as 0.1 kPa, is the isentropic efficiency, taken as 0.8, is the outlet pressure, is the air mass flow.

3. The method of claim 1, wherein, In S3, the global variables of the Harris hawk algorithm include the number of Harris hawks , dimensions , upper limit of variable interval and lower limit , maximum number of iterations ; Wherein, each Harris hawk corresponds to an air compression system control strategy, and the air compression system control strategy comprises the air tank output pressure and the air compressor start and stop time; In the process of initializing global variables of the improved Harris hawk algorithm, the air compressor output pressure is taken as an optimization variable: wherein, is an optimization variable vector, denotes a gas tank, denotes an output pressure of the gas tank, denotes an air compressor, denotes an on time of the air compressor, denotes an off time of the air compressor, denotes an output pressure of the air compressor; wherein , As a time variable, the following decimal manner is adopted: wherein , represents hours, , represents minutes.

4. The method of claim 1, wherein, In S4, the process of optimizing population initialization of the original Harris hawk through reverse mapping comprises: S4-1, number of Harris eagles , dimension , upper limit of variable interval and lower limit Optimization population is randomly generated as shown in equation (4): wherein, is a randomly generated optimized population, is a randomly generated random number in the interval 0 to 1 following a uniform distribution, is a randomly generated matrix of random numbers with rows and columns. S4-2, calculate The fitness value of each individual, i.e. the total energy consumption produced by each air compression system control strategy, is calculated as shown in equation (5): in, This is the function for calculating the total energy consumption of the compressed air system, i.e., the fitness calculation function. for The Control strategy, For random population fitness, For the first The total energy consumption of the control strategy, i.e. the first The fitness value of the control strategy; The reverse population is generated according to formula (6), and the fitness value of the reverse population is calculated according to formula (7), and formula (6) is as follows: wherein, is the generated reverse population; Formula (7) is as follows: wherein, is the fitness of the reverse population, is the fitness of the reverse population, is the fitness of the reverse population. S4-3, fitness of random population and reverse population fitness merged into , fitness of random population and reverse population merged into , indicates that the reverse population is longitudinally spliced after the random population, and the total energy consumption of is sorted, and the control strategy with the lowest total energy consumption is selected to be combined into the initialization population to be iterated . .

5. The method of claim 3, wherein, In S5, the process of iteratively optimizing the initialized Harris hawk population comprises: The value calculated by formula (8) The size of the value selects different evolution strategies: wherein is the current iteration number, is a is a random number within.

6. The method of claim 5, wherein, In S5, the process of selecting different evolution strategies comprises: When the absolute value of the difference between the two is greater than or equal to 1, the evolutionary strategy shown in the individual selection formula (9) at this time: wherein, is the evolutionary offspring, is the current best individual, is the current individual, is the natural constant, is a random number within, is the average position of the current population, is the absolute distance of the current individual from the best individual, and the calculation formula is as follows: 。 7. The method of claim 6, wherein, In S5, the process of selecting different evolution strategies comprises: When Less than 1, a development strategy is executed, which includes four specific strategies, by one Random decimal With The size of the value to determine which evolutionary strategy to execute: (1) greater than or equal to 0.5, and an absolute value of the difference between the two values is greater than or equal to 0.5, at which time an evolutionary strategy shown in Equation (10) is executed. wherein, is a random number within the interval, is the distance of the current best individual and the current individual; (2) greater than or equal to 0.5, and an absolute value less than 0.5, at which time an evolutionary strategy shown in equation (11) is executed. (3) less than 0.5, and greater than or equal to 0.5, in which case the evolutionary strategy shown in equation (12) is performed. wherein is a random number between 0 and 1. (4) less than 0.5, and less than 0.5, in which case an evolutionary strategy as shown in equation (13) is performed. wherein is a random number between 0 and 1.

8. The method of claim 1, wherein, In S6, the process of executing the partition elitism strategy under variable isolation comprises: For the output pressure of the air tank, a disturbance strategy based on Lévy flight is adopted.

9. The method of claim 8, wherein, In S6, the process of executing the partition elitism strategy under variable isolation further comprises: For the output pressure and the start and stop time of the air compressor, a differential evolution strategy is adopted.

Images