Air compression station compressed air flow distribution method, device and electronic equipment

Abstract

The application provides a compressed air flow distribution method and device of an air compression station and electronic equipment, and belongs to the technical field of industrial automation. The method comprises the following steps: obtaining operation data of air compression station component equipment, terminal equipment gas source demand and at least one control target; using a multi-objective optimization algorithm; using a deep learning network to analyze the operation data according to the control target; using a weighted sum method to predict optimal operation parameters and predicted operation data of the air compression station component equipment when the terminal equipment gas source demand is met; controlling the operation of the air compression station component equipment according to the updated optimal operation parameters; and dynamically distributing compressed air flow to the terminal. Through the method, the degree of automation is improved, dynamic distribution and optimization of the compressed air flow of the air compression station can be realized, the distribution efficiency of the gas source and efficient scheduling of the unit are improved, and production interruption caused by improper flow distribution is effectively avoided.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation technology and relates to a method, device and electronic equipment for distributing compressed air flow in an air compressor station. Background Technology

[0002] With the continuous development of industrial automation and intelligent equipment, traditional air compressor stations face many challenges in flow distribution, monitoring, and management. Particularly in flow distribution, existing methods are typically based on fixed rules, making dynamic adjustments based on real-time demand difficult, leading to energy waste or insufficient air supply. Furthermore, control relies on individual air compressor systems for single-unit control, hindering coordinated scheduling among multiple units. This further exacerbates the problem when operating conditions change, preventing the existing system from quickly adjusting flow distribution, resulting in low operating efficiency and insufficient support for overall energy efficiency optimization of the air compressor station, making it difficult to effectively reduce energy consumption. In short, existing technologies struggle to achieve efficient and precise flow control and optimization for air compressor stations. Summary of the Invention

[0003] The purpose of this invention is to address the aforementioned problems in existing technologies by proposing a method, apparatus, and electronic equipment for distributing compressed air flow in an air compressor station.

[0004] The objective of this invention can be achieved through the following technical solution: a method for distributing compressed air flow in an air compressor station, comprising:

[0005] Acquire operational data of the components of the air compressor station, air supply requirements of terminal equipment, and at least one control objective;

[0006] A multi-objective optimization algorithm is adopted. Based on the control objective, a deep learning network is used to analyze the operating data. The weighted sum method is used to predict the optimal operating parameters and predicted operating data of the air compressor station components when the gas supply demand of the terminal equipment is met.

[0007] The operation of the air compressor station components is controlled according to the optimal operating parameters to obtain real-time operating data. The real-time operating data is analyzed based on the predicted operating data to apply rewards or penalties. The deep learning network dynamically updates its weights based on the rewards or penalties.

[0008] The optimal operating parameters of the air compressor station components are updated based on the updated weight predictions, and the operation of the air compressor station components is controlled based on the updated optimal operating parameters to dynamically allocate compressed air flow to the terminals.

[0009] As an optional embodiment of the present invention, a multi-objective optimization algorithm is employed. Based on the control objective, a deep learning network is used to analyze the operating data, and a weighted sum method is used to predict the optimal operating parameters of the air compressor station components when meeting the gas supply demand of the terminal equipment. This includes:

[0010] The control objective includes minimizing the total energy consumption of the air compressor station while meeting the air supply requirements of the terminal equipment.

[0011] Obtain the relationship between the power of each compressor and the corresponding output compressed air flow rate;

[0012] Based on the control objective, the optimal operating time and optimal air supply of each compressor are predicted by weighted calculation using the relationship and the relationship for minimizing the total energy consumption of the air compressor station, while satisfying the air source demand of the terminal equipment and minimizing the total energy consumption of the air compressor station.

[0013] As an optional embodiment of the present invention, the control objective further includes: the optimal gas supply is greater than or equal to the gas source demand of the terminal equipment.

[0014] As an optional embodiment of the present invention, a multi-objective optimization algorithm is employed. Based on the control objective, a deep learning network is used to analyze the operational data. A weighted sum method is then used to predict the optimal operating parameters of the air compressor station components when meeting the gas supply demand of the terminal equipment. The method further includes:

[0015] The control objective also includes minimizing the pressure fluctuation deviation of the air storage tank when the pressure in the air compressor station meets the compressed air supply process.

[0016] A dynamic model of pressure fluctuation is constructed based on the gas storage tank volume, inlet flow rate, and outlet flow rate.

[0017] Based on the control objective, when the pressure inside the gas storage tank fluctuates, the minimum pressure fluctuation deviation is predicted by weighted calculation based on the pressure fluctuation dynamic model, when the gas supply demand of the terminal equipment is met and the pressure of the gas storage tank is within the preset pressure range.

[0018] As an optional embodiment of the present invention, the deep learning network includes:

[0019] The control objective is input into the state space of the deep learning network;

[0020] The weights corresponding to the control target are input into the dynamic space of the deep learning network;

[0021] Based on the state space and dynamic space, a corresponding reward function is constructed, and based on the state space, the dynamic space, and the reward function, a deep learning network is constructed to determine the optimal operating parameters of the air compressor station's components.

[0022] As an optional embodiment of the present invention, it further includes:

[0023] All historical optimal operating parameters and corresponding historical operating data within a preset time range are acquired at a fixed frequency. Global strategy correction is performed on the historical optimal operating parameters and the historical operating data to obtain the optimal operating parameter set and the corresponding operating dataset.

[0024] The optimal operating parameter set and the operating dataset are input into the deep learning network and trained. When new operating data is received, the trained deep learning network predicts the optimal operating parameters and predicted operating data of the air compressor station components when the air source demand of the terminal equipment is met.

[0025] As an optional embodiment of the present invention, a global strategy correction is performed on the historical optimal operating parameters and the historical operating data to obtain an optimal operating parameter set and a corresponding operating dataset, including:

[0026] The historical optimal operating parameters and the corresponding historical operating data are used as a set of data to analyze whether each set of data meets the control objective.

[0027] The fitness of multiple sets of data that meet the control objective is sorted from high to low. A preset number of sets of data are obtained from the sorting results and integrated into the optimal set of operating parameters and the corresponding operating dataset.

[0028] As an optional embodiment of the present invention, real-time operational data is analyzed based on the predicted operational data to apply rewards or penalties, and the deep learning network dynamically updates its weights based on the rewards or penalties; including:

[0029] Analyze the predicted and real-time operating data of the control target. If the analysis results meet the preset rules, then apply the corresponding reward or penalty, and update the weight of the control target based on the reward or penalty.

[0030] The present invention also provides a compressed air flow distribution device for an air compressor station, comprising:

[0031] The data acquisition module is used to acquire the operating data of the components of the air compressor station, the air source demand of the terminal equipment, and at least one control target.

[0032] The data analysis module is used to employ a multi-objective optimization algorithm to analyze the operating data using a deep learning network based on the control objective, and to predict the optimal operating parameters and predicted operating data of the air compressor station components when meeting the gas source demand of the terminal equipment using a weighted sum method.

[0033] The feedback module is used to control the operation of the air compressor station components according to the optimal operating parameters, obtain real-time operating data, analyze the real-time operating data according to the predicted operating data, and give rewards or penalties. The deep learning network dynamically updates its weights according to the rewards or penalties.

[0034] The flow distribution module is used to predict and update the optimal operating parameters of the air compressor station components based on the updated weights, control the operation of the air compressor station components based on the updated optimal operating parameters, and dynamically distribute compressed air flow to the terminal.

[0035] The present invention also provides an electronic device, comprising:

[0036] processor;

[0037] Memory used to store processor-executable instructions;

[0038] The processor is configured to implement the compressed air flow distribution method of the air compressor station as described above when executing the executable instructions.

[0039] Compared with existing technologies, this invention uses a multi-objective optimization algorithm combined with a deep learning network to predict the optimal operating parameters, thereby improving the degree of automation. It can realize the dynamic allocation and optimization of compressed air flow in the air compressor station, improve the allocation efficiency of air source and the efficient scheduling of units, improve the response speed and stability of the air compressor station system, and effectively avoid production interruptions caused by improper flow allocation. Attached Figure Description

[0040] Figure 1 This is a flowchart of the compressed air flow distribution method of the air compressor station according to an embodiment of the present invention;

[0041] Figure 2 This is a block diagram of the compressed air flow distribution device of the air compressor station according to an embodiment of the present invention. Detailed Implementation

[0042] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0043] Example 1

[0044] An air compressor station, also known as an air compression station, is a system used in industrial settings to provide compressed air. Its core components include an air compressor, air tank, post-processing equipment, piping system, and corresponding control and monitoring equipment. Compressed air is an important form of energy in many industrial sectors and is widely used to drive equipment and control pneumatic systems.

[0045] Flow distribution refers to the process of allocating compressed air to different usage ports or production lines according to demand. In an air compressor station, flow distribution needs to meet the real-time needs of different equipment while optimizing the overall system's energy efficiency.

[0046] With the continuous development of industrial automation and intelligent equipment, this embodiment proposes a method for distributing compressed air flow in an air compressor station, aiming to solve the technical problems reflected in the background technology, such as... Figure 1 As shown, a method for distributing compressed air flow in an air compressor station includes:

[0047] S1, acquire the operating data of the components of the air compressor station, the air source demand of the terminal equipment, and at least one control target;

[0048] S2, using a multi-objective optimization algorithm, a deep learning network is used to analyze the operating data based on the control objective, and a weighted sum method is used to predict the optimal operating parameters and predicted operating data of the air compressor station components when the gas source demand of the terminal equipment is met;

[0049] S3, control the operation of the air compressor station components according to the optimal operating parameters, obtain real-time operating data, analyze the real-time operating data according to the predicted operating data, and give rewards or penalties. The deep learning network dynamically updates its weights according to the rewards or penalties.

[0050] S4, based on the updated weights, predict and update the optimal operating parameters of the air compressor station components, control the operation of the air compressor station components based on the updated optimal operating parameters, and dynamically allocate compressed air flow to the terminal.

[0051] Sensors installed on various components of the air compressor station, such as flow sensors, pressure sensors, and temperature sensors, collect real-time operating parameters and corresponding data. The collected data then undergoes preprocessing and feature extraction, including time-series alignment, noise filtering, and outlier repair. Real-time data also includes the air demand of the terminal equipment. Control objectives are set by the operators; for example, minimizing system energy consumption and ensuring stable compressed air reserves. In other words, while meeting the air demand of the terminal equipment, the aforementioned control objectives must also be considered.

[0052] The number of control objectives may be one or more. In this embodiment, a deep learning agent is used to analyze operational data by comprehensively considering the control objectives through a built-in multi-objective optimization algorithm. Weights are initialized for each control objective, and a weighted sum method is used to predict the optimal action of the component equipment, i.e., the optimal operating parameters of the component equipment. Predicted operational data is also recorded when the component equipment operates according to the optimal operating parameters. The operating parameters of the component equipment are adjusted to the predicted optimal operating parameters, and the corresponding real-time operational data is recorded. This embodiment also designs a dynamic weight adjustment strategy based on DQN. Based on the aforementioned predicted operational data, real-time operational data is analyzed to evaluate the performance of the air compressor station's component equipment, such as energy efficiency, gas source demand satisfaction, and gas tank pressure stability. A reward or penalty is given based on the performance through a reward function to guide the next adjustment of the weights of each control objective for the component equipment.

[0053] After weight adjustment, the optimal operating parameters of the constituent equipment are updated accordingly. The constituent equipment operates according to the updated optimal operating parameters, dynamically allocating compressed air flow to the terminal. By employing a deep learning agent to analyze the current operating data of the air compressor station's constituent equipment, and based on the control objective, predicting the next optimal operating parameters and corresponding predicted operating data of the air compressor station's constituent equipment, for each set of constituent equipment's operating data, the latest predicted operating data is analyzed, and rewards or penalties are given. Based on the rewards or penalties, the weights corresponding to each control objective are adjusted, thereby obtaining the corresponding optimal operating parameters. This avoids the limitation of fixed weights in the traditional weighted sum method. The operating data obtained based on the optimal operating parameters satisfies the air supply demand of the terminal equipment while also meeting the control objective. This process is repeated, and over time, the deep learning agent, utilizing its convergence characteristics, gradually adapts to the optimization of compressed air flow allocation under various operating conditions.

[0054] Preferably, a multi-objective optimization algorithm is employed. Based on the control objective, a deep learning network is used to analyze the operating data. A weighted sum method is used to predict the optimal operating parameters of the air compressor station components when meeting the air supply demand of the terminal equipment. This includes: the control objective includes minimizing the total energy consumption of the air compressor station when meeting the air supply demand of the terminal equipment; obtaining the relationship between the power of each compressor and the corresponding output compressed air flow rate; and based on the control objective, using the relationship and the relationship for minimizing the total energy consumption of the air compressor station, a weighted prediction is made of the optimal operating time and optimal air supply of each compressor when the air supply demand of the terminal equipment is met and the total energy consumption of the air compressor station is minimized.

[0055] In this embodiment, one of the control objectives is to minimize the energy consumption of the air compressor station, that is, to minimize the energy consumption of the air compressor station while meeting the air supply requirements of all terminal equipment. This is expressed by the following formula:

[0056]

[0057] Among them, E Total It is the total energy consumption of the air compressor station; P i P is the power of the i-th air compressor; i With output flow rate Q i The relationship is non-linear and can be fitted using the characteristic curve of an air compressor:

[0058]

[0059] In the formula: a, b, and c are constants, generated by the i-th air compressor through multiple power P groups. i With output flow rate Q i The experiment yielded the following specific results: the air compressor characteristic curve fitting process involves obtaining the air compressor's power P during operation. i and output flow Q i A relationship diagram or scatter plot showing the power P i and output flow Q i Fit to a quadratic function. Output flow rate Q i That is, the actual gas supply of the i-th terminal device, in m³. 3 / min; T i Let be the operating time of the i-th air compressor. For this control objective, a weighted sum of this expression can be performed to optimize the subsequent start-stop states T of each air compressor. i and traffic Q i This reduces the total energy consumption of the air compressor station. The optimized T i That is, the optimal running time, Q. i The optimal air supply means that when the air compressor runs for the optimal time and the actual air supply is the optimal air supply, the energy consumption of the air compressor station is minimized.

[0060] Preferably, the control objective further includes: the optimal gas supply is greater than or equal to the gas source demand of the terminal equipment.

[0061] Each terminal device in an air compressor station, such as a production line or pneumatic tools, has specific air supply requirements. The control objective is to ensure that the requirements of each terminal are met. This constraint can be expressed by the following inequality:

[0062]

[0063] D i Let Q be the air supply demand of the i-th terminal device, and m be the total number of terminal devices. Considering only the relationship between the air supply and the air supply demand of the terminal devices, when the air supply equals the air supply demand of the terminal devices, it perfectly meets the terminal needs and does not result in wasted compressed air flow. However, due to the need to consider the system energy consumption of the air compressor station and the pressure inside the air storage tank, Q...i The minimum system energy consumption is not necessarily determined by the gas demand of the terminal equipment. As mentioned above, system energy consumption is also related to the operating time T. i This is relevant. Therefore, the control objective aims to ensure that when the air compressor station distributes flow, the optimized Q value for each terminal device is achieved. i All of them can meet the air supply requirements of each terminal device, and will not cause production interruption due to insufficient supply. In addition, the air storage tank maintains a certain pressure range during the distribution of compressed air flow.

[0064] Preferably, a multi-objective optimization algorithm is employed, which uses a deep learning network to analyze the operating data based on the control objective, and uses a weighted sum method to predict the optimal operating parameters of the air compressor station components when meeting the gas supply demand of the terminal equipment. The method further includes:

[0065] The control objectives also include minimizing the pressure fluctuation deviation of the air storage tank when the pressure in the air compressor station meets the compressed air supply process; constructing a dynamic pressure fluctuation model based on the air storage tank volume, inlet flow rate, and outlet flow rate; and based on the control objectives, when the pressure inside the air storage tank fluctuates, using the dynamic pressure fluctuation model to predict the minimum pressure fluctuation deviation when the air source demand of the terminal equipment is met and the pressure of the air storage tank is within a preset pressure range.

[0066] The air compressor station's storage tank needs to maintain a certain pressure range during the compressed air supply process. Too low a pressure will lead to insufficient air supply, while too high a pressure will waste energy and increase equipment load. This can be represented by the following constraints:

[0067] P min ≤P 储罐 ≤P max

[0068] Pressure fluctuations are mainly caused by sudden load changes and compressor response lag, and their dynamic model can be simplified as follows:

[0069]

[0070] In the formula, V is the tank volume, in m³. 3 Q in Q out These are the intake flow rate and the exhaust flow rate, respectively, in m³. 3 / min; The control objective is to optimize and minimize the pressure deviation:

[0071] min P 储罐 -Pave

[0072] Among them, P aveThe system average pressure can be obtained using conventional techniques in this field and will not be elaborated upon here. Regarding the aforementioned control objectives, when allocating compressed air flow, these objectives are comprehensively considered, transforming multi-objective optimization into single-objective optimization, and a weighted sum method is used for processing:

[0073]

[0074] In the formula, ω1, ω2, and ω3 are the weights of each control objective, and E Total This is the total energy consumption of the air compressor station. Indicates the degree to which gas supply demand is met; |P 储罐 -P ave | represents the penalty term for pressure fluctuations in the air storage tank. The resulting F is the optimal operating parameters and predicted operating data of the components of the air compressor station when meeting the air source demand of the terminal equipment. In the scenario of compressed air flow distribution in the air compressor station, based on the operating parameters of each compressor, the operating data of the user end and intermediate transmission and processing links, and considering all control objectives, dynamic optimization of the optimal operating parameters and predicted operating data is performed. It should be noted that, apart from the control objectives in this embodiment, operators can also add or change control objectives in actual applications. However, regardless of the number of control objectives, they can all be transformed into single-objective optimization through a multi-objective optimization algorithm, and the optimal operating parameters and predicted operating data of the corresponding components of the equipment under the condition of meeting the air source demand of the terminal equipment can be obtained using a weighted sum method.

[0075] Preferably, the deep learning network includes:

[0076] The control objective is input into the state space of the deep learning network; the weights corresponding to the control objective are input into the dynamic space of the deep learning network; a corresponding reward function is constructed based on the state space and the dynamic space; and a deep learning network for the optimal operating parameters of the air compressor station components is constructed based on the state space, the dynamic space, and the reward function.

[0077] In this embodiment, the deep learning network is mainly a deep Q-network, and a dynamic weight adjustment strategy based on DQN is designed. Specifically, the target input is controlled to enter the state space S. t In the middle, that is, the pressure P of the gas storage tank 储罐 Power P of each air compressor i Gas demand and total energy consumption of terminal equipment (E) Total Action space A t Let ω1, ω2, and ω3 be the weight vectors, used to adjust the weight vectors within a certain range. The reward function reflects the combined optimization effect of multiple objectives and is defined as:

[0078]

[0079] In the formula: λ is the stability reward coefficient, used to amplify the positive incentive brought by long-term stability (e.g., λ = 1); Stability_Bonus is, for example, an additional reward that can be dynamically calculated based on the pressure fluctuation of the storage tank, such as an additional reward of +10 when the standard deviation of pressure fluctuation is below the threshold for 10 consecutive minutes. Furthermore, the deep learning network used to predict the optimal operating parameters of the constituent equipment and predict operating data is:

[0080]

[0081] S t As the current state, A t For the current sampling action, F t α is the reward value, α is the learning rate, and γ is the discount factor, which measures the impact of future rewards.

[0082] Deep Q-networks help air compressor station systems make optimal decisions in complex, multi-objective constrained environments, improving operating efficiency, reducing energy consumption, and ensuring system stability.

[0083] Preferably, it further includes:

[0084] All historical optimal operating parameters and corresponding historical operating data within a preset time range are acquired at a fixed frequency. Global strategy correction is performed on the historical optimal operating parameters and the historical operating data to obtain the optimal operating parameter set and the corresponding operating dataset. The optimal operating parameter set and the operating dataset are input into the deep learning network and trained. When receiving new operating data, the trained deep learning network predicts the optimal operating parameters and predicted operating data of the air compressor station components when meeting the gas source demand of the terminal equipment.

[0085] In this embodiment, the historical optimal operating parameters for a specific time period are globally revised at a fixed frequency of once a week, resulting in the optimal operating parameter set and corresponding operating dataset for that time period. This dataset is then used to train a deep learning network. The deep learning network utilizes its self-learning capabilities to learn the features of the optimal operating parameter set and the corresponding operating dataset, continuously iterating the strategy. After training, the deep learning network becomes increasingly accurate in predicting the optimal operating parameters and operating data of the air compressor station's components when receiving real-time operating data; in other words, the deep learning network undergoes periodic optimization. It should be noted that the preset time range is set by the user based on the learning progress of the deep learning agent and is not limited here.

[0086] Preferably, a global strategy correction is performed on the historical optimal operating parameters and the historical operating data to obtain the optimal operating parameter set and the corresponding operating dataset, including:

[0087] The historical optimal operating parameters and the corresponding historical operating data are taken as a set of data, and each set of data is analyzed to see if it meets the control objective. The fitness of multiple sets of data that meet the control objective is sorted from high to low. A preset number of multiple sets of data are obtained from the sorting results and integrated into the optimal operating parameter set and the corresponding operating dataset.

[0088] When predicting optimal operating parameters, corresponding predicted operating data is generated. Then, based on whether real-time operating data meets the control objectives, rewards or penalties are given based on the analysis of the predicted operating data and real-time operating data. The overall strategy correction aims to improve the performance of the air compressor station. Therefore, when obtaining historical optimal operating parameters and corresponding historical operating data, these can be treated as a set of data. Each set of data is analyzed sequentially to determine if it meets the control objectives. If not, it is ignored; if it does, it is retained. For the retained sets of data, the results F corresponding to each set are ranked by fitness. That is, among the multiple sets of historical optimal operating parameters and corresponding historical operating data that meet the control objectives, they are sorted from high to low fitness, and a predetermined number of sets of historical optimal operating parameters and corresponding historical operating data are selected and integrated into an optimal operating parameter set and a corresponding operating dataset.

[0089] Preferably, real-time operational data is analyzed based on the predicted operational data to apply rewards or penalties, and the deep learning network dynamically updates its weights based on the rewards or penalties; including:

[0090] Analyze the predicted and real-time operating data of the control target. If the analysis results meet the preset rules, then apply the corresponding reward or penalty, and update the weight of the control target based on the reward or penalty.

[0091] The example above illustrates an additional reward that can be dynamically calculated based on tank pressure fluctuations. For instance, if the standard deviation of pressure fluctuation is below a threshold for 10 consecutive minutes, an additional reward of +10 is given. In practical applications, users can also set other rules to compare and analyze predicted and real-time operating data. For example, in real-time operating data, the total energy consumption E... total Compared to the total energy consumption E in the predicted operating data total The lower value indicates that the actual total system energy consumption is lower than predicted, which allows for a more accurate assessment of the total energy consumption E. total One aspect involves rewarding or penalizing the device. Each time the device performs an action based on the predicted optimal operating parameters, the deep learning network will reward or penalize based on the analysis results, and further adjust the optimal operating parameters for the next operation based on the reward or penalty. Over time, the strategy of the intelligent multi-objective optimization algorithm will gradually converge and adapt to the optimal allocation of compressed air flow under various operating conditions.

[0092] When analyzing real-time operational data based on predicted operational data, the health status of corresponding equipment can be further determined, and potential faults can be predicted in advance. Remote control is also possible, allowing administrators to remotely start and stop equipment, modify parameters, or schedule tasks through the platform, significantly reducing manual intervention and control of the air compressor station system.

[0093] For the computational functions involved, edge computing can be used to distribute some of the computational functions of the deep learning network to the device, thereby reducing the burden on the central server in the air compressor station.

[0094] By employing the above methods and combining multi-objective optimization algorithms with deep learning networks to predict optimal operating parameters, the level of automation is improved. This enables dynamic allocation and optimization of compressed air flow in air compressor stations, enhances the efficiency of air source allocation and efficient unit scheduling, improves the response speed and stability of the air compressor station system, and effectively avoids production interruptions caused by improper flow allocation.

[0095] Example 2

[0096] Based on the same principle as the aforementioned method, a compressed air flow distribution device 100 for an air compressor station is also proposed, such as... Figure 2 As shown, it includes:

[0097] The data acquisition module 110 is used to acquire the operating data of the components of the air compressor station, the air source demand of the terminal equipment, and at least one control target.

[0098] The data analysis module 120 is used to use a multi-objective optimization algorithm to analyze the operating data using a deep learning network based on the control objective, and to use a weighted sum method to predict the optimal operating parameters and predicted operating data of the air compressor station components when meeting the gas source demand of the terminal equipment.

[0099] Feedback module 130 is used to control the operation of the air compressor station components according to the optimal operating parameters, obtain real-time operating data, analyze the real-time operating data according to the predicted operating data, and give rewards or penalties. The deep learning network dynamically updates its weights according to the rewards or penalties.

[0100] The flow distribution module 140 is used to predict and update the optimal operating parameters of the air compressor station components based on the updated weights, control the operation of the air compressor station components based on the updated optimal operating parameters, and dynamically distribute compressed air flow to the terminal.

[0101] Example 3

[0102] Furthermore, an electronic device is proposed, comprising:

[0103] processor;

[0104] Memory used to store processor-executable instructions;

[0105] The processor is configured to implement the compressed air flow distribution method of the air compressor station described in Embodiment 1 when executing the executable instructions.

[0106] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0107] Furthermore, it should be noted that the use of terms such as "first," "second," and "a" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified. The terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two elements or the interaction between two elements, unless otherwise explicitly specified. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0108] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0109] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A method of distributing compressed air flow in an air compression station, characterized in that, include: Acquire operational data of the components of the air compressor station, the air supply demand of the terminal equipment, and at least one control objective; A multi-objective optimization algorithm is adopted. Based on the control objective, a deep learning network is used to analyze the operating data. The weighted sum method is used to predict the optimal operating parameters and predicted operating data of the air compressor station components when the gas supply demand of the terminal equipment is met. The operation of the air compressor station components is controlled according to the optimal operating parameters to obtain real-time operating data. The real-time operating data is analyzed based on the predicted operating data to apply rewards or penalties. The deep learning network dynamically updates its weights based on the rewards or penalties. The optimal operating parameters of the air compressor station components are updated based on the updated weight predictions, and the operation of the air compressor station components is controlled based on the updated optimal operating parameters to dynamically allocate compressed air flow to the terminals.

2. A method of distributing compressed air flow in an air compression station according to claim 1, characterized in that, A multi-objective optimization algorithm is employed. Based on the control objective, a deep learning network is used to analyze the operational data. A weighted sum method is then used to predict the optimal operating parameters of the air compressor station's components to meet the gas supply demand of the terminal equipment. These parameters include: The control objective includes minimizing the total energy consumption of the air compressor station while meeting the air supply requirements of the terminal equipment. Obtain the relationship between the power of each compressor and the corresponding output compressed air flow rate; Based on the control objective, the optimal operating time and optimal air supply of each compressor are predicted by weighted calculation using the relationship and the relationship for minimizing the total energy consumption of the air compressor station, while satisfying the air source demand of the terminal equipment and minimizing the total energy consumption of the air compressor station.

3. The compressed air flow distribution method for an air compressor station according to claim 2, characterized in that, The control objective also includes: the optimal gas supply is greater than or equal to the gas demand of the terminal equipment.

4. The compressed air flow distribution method for an air compressor station according to claim 2, characterized in that, A multi-objective optimization algorithm is employed, which uses a deep learning network to analyze the operating data based on the control objective. A weighted sum method is used to predict the optimal operating parameters of the air compressor station components when meeting the gas supply demand of the terminal equipment. The algorithm also includes: The control objective also includes minimizing the pressure fluctuation deviation of the air storage tank when the pressure in the air compressor station meets the compressed air supply process. A dynamic model of pressure fluctuation is constructed based on the gas storage tank volume, inlet flow rate, and outlet flow rate. Based on the control objective, when the pressure inside the gas storage tank fluctuates, the minimum pressure fluctuation deviation is predicted by weighted calculation based on the pressure fluctuation dynamic model, when the gas supply demand of the terminal equipment is met and the pressure of the gas storage tank is within the preset pressure range.

5. The compressed air flow distribution method for an air compressor station according to claim 1, characterized in that, The deep learning network includes: The control objective is input into the state space of the deep learning network; The weights corresponding to the control target are input into the dynamic space of the deep learning network; Based on the state space and dynamic space, a corresponding reward function is constructed, and based on the state space, the dynamic space, and the reward function, a deep learning network is constructed to determine the optimal operating parameters of the air compressor station's components.

6. The compressed air flow distribution method for an air compressor station according to claim 5, characterized in that, Also includes: All historical optimal operating parameters and corresponding historical operating data within a preset time range are acquired at a fixed frequency. Global strategy correction is performed on the historical optimal operating parameters and the historical operating data to obtain the optimal operating parameter set and the corresponding operating dataset. The optimal operating parameter set and the operating dataset are input into the deep learning network and trained. When new operating data is received, the trained deep learning network predicts the optimal operating parameters and predicted operating data of the air compressor station components when the air source demand of the terminal equipment is met.

7. The compressed air flow distribution method for an air compressor station according to claim 6, characterized in that, A global strategy correction is performed on the historical optimal operating parameters and the historical operating data to obtain the optimal operating parameter set and the corresponding operating dataset, including: The historical optimal operating parameters and the corresponding historical operating data are used as a set of data to analyze whether each set of data meets the control objective. The fitness of multiple sets of data that meet the control objective is sorted from high to low. A preset number of sets of data are obtained from the sorting results and integrated into the optimal set of operating parameters and the corresponding operating dataset.

8. The compressed air flow distribution method for an air compressor station according to claim 6, characterized in that, Based on the predicted running data, real-time running data is analyzed, and rewards or penalties are applied. The deep learning network dynamically updates its weights based on the rewards or penalties. include: Analyze the predicted and real-time operating data of the control target. If the analysis results meet the preset rules, then apply the corresponding reward or penalty, and update the weight of the control target based on the reward or penalty.

9. A compressed air flow distribution device for an air compressor station, characterized in that, include: The data acquisition module is used to acquire the operating data of the components of the air compressor station, the air source demand of the terminal equipment, and at least one control target. The data analysis module is used to employ a multi-objective optimization algorithm to analyze the operating data using a deep learning network based on the control objective, and to predict the optimal operating parameters and predicted operating data of the air compressor station components when meeting the gas source demand of the terminal equipment using a weighted sum method. The feedback module is used to control the operation of the air compressor station components according to the optimal operating parameters, obtain real-time operating data, analyze the real-time operating data according to the predicted operating data, and give rewards or penalties. The deep learning network dynamically updates its weights according to the rewards or penalties. The flow distribution module is used to predict and update the optimal operating parameters of the air compressor station components based on the updated weights, control the operation of the air compressor station components based on the updated optimal operating parameters, and dynamically distribute compressed air flow to the terminal.

10. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to implement the compressed air flow distribution method of the air compressor station according to any one of claims 1-8 when executing the executable instructions.

Images