Intelligent load control method for modular air compressors based on Internet of Things (IoT) communication

Abstract

This invention discloses a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication, relating to the field of air compressor control technology. The method includes: collecting air compressor operating data and forming control input data; processing the operating data through a causal convolutional network to generate initial candidate control solutions; optimizing the initial candidate control solutions and determining the target candidate control solutions; and executing the target candidate control solutions and performing rolling updates.

Description

Technical Field

[0001] This invention relates to the field of air compressor control technology, specifically to a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication. Background Technology

[0002] Air compressor systems are widely used in industrial settings such as manufacturing, processing, transportation, and automated production to continuously supply compressed air to the air-consuming sites. With faster production cycles and greater fluctuations in air load, traditional single-unit or multi-unit parallel air compressor supply regulation methods are no longer sufficient to balance supply stability, operational efficiency, and timely control. This is especially true in scenarios where multiple modular operating units operate collaboratively; existing methods typically lack mechanisms for comparing and selecting multiple alternative control schemes, making it difficult to simultaneously address supply matching, switching frequency control, and overall operational efficiency.

[0003] Chinese invention patent application CN120777196A, published on October 14, 2025, describes a collaborative control architecture that incorporates a first pneumatic adjustment module, a second pneumatic adjustment module, a temperature and pressure monitoring module, a PLC control module, a graded adjustment module, and an opening adjustment module. This architecture enables precise and flexible control of pressure and temperature between the first and second stages. It addresses the problem of excessively high first-stage exhaust temperature in oil-free screw air compressors under low-speed loads, which limits the minimum energy-saving speed. This allows the unit to operate safely at lower speeds and improves energy efficiency.

[0004] However, the technical solutions in the aforementioned comparative documents mainly focus on the interstage pressure and temperature control during the first and second stage compression processes of oil-free screw air compressors. Their core lies in the coordinated adjustment of flexible pipelines, valve opening, and pressure and temperature corrections. They primarily address the issues of fixed interstage parameters within a single air compressor, lag in thermal inertia response, and insufficient real-time control of compression efficiency. This solution does not establish an intelligent load control mechanism for the coordinated operation of multiple module operating units under IoT communication conditions, nor does it construct a predictive model for the continuous temporal changes in total air consumption and pipeline pressure. Furthermore, it does not provide a comparison and selection process based on multiple alternative control schemes. Therefore, in the scenario of a modular air compressor system, it is still difficult to simultaneously consider subsequent air supply demand prediction, load distribution of multiple module operating units, and start-stop switching control.

[0005] To this end, the present invention provides a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication. By predicting subsequent air supply demand based on total air flow data, pipeline pressure data, and output flow data of each module operating unit within a continuous unit time, this method generates and optimizes multiple candidate control schemes, thereby improving the air supply matching, load allocation rationality, and overall operating efficiency of the modular air compressor system.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication, comprising: S1. Collect operational data and generate control input data; S2. Generate initial candidate control solutions based on the control input data using a causal convolutional network; S3. Optimize the initial candidate control solution and determine the target candidate control solution; S4. Execute the target candidate control solution and perform rolling updates.

[0008] Preferably, the control input data of the main modular air compressor system and multiple historical control input data are periodically acquired through the IoT communication layer and sent to the load control center. The control input data and multiple historical control input data include: output flow data, operating frequency data, status flag data, loading status data, unloading status data, and start / stop status data periodically collected by each module operating unit, as well as the total air consumption data of the main pipeline. Among them, the output flow data is acquired by the flow sensor installed on the outlet pipeline of the corresponding module operating unit, the operating frequency data is acquired by the frequency converter feedback signal of the corresponding module operating unit, the loading status data, unloading status data, and start / stop status data are acquired by the controller status feedback signal of the corresponding module operating unit, and the total air consumption data is acquired by the flow sensor installed on the main pipeline.

[0009] Preferably, the load control center inputs the control input data and multiple historical control input data into a causal convolutional network. The causal convolutional network has the function of extracting temporal correlation features, and predicts the next x units of time as prediction units. At each prediction unit, the causal convolutional network outputs the target gas supply demand value, the initial target output flow value of each module operating unit, the target status marker value, the start-up time, the shutdown time, the loading switching time, the unloading switching time, the input order, and the exit order. The target status marker value refers to the discrete value used to characterize the target operating state of each module operating unit at each prediction unit. For any module operating unit at any prediction unit, the target status marker value is 0, 1, or 2. In the target status marker value, 0 indicates shutdown, 1 indicates unloading operation, and 2 indicates loading operation. Where x is a positive integer greater than 1, which can be adjusted according to the actual situation; The causal convolutional network receives regulatory input data and multiple historical regulatory input data. It performs convolution operations on the regulatory input data and multiple historical regulatory input data in chronological order to extract the temporal relationship between gas consumption data and output flow data. Based on this temporal relationship, the causal convolutional network outputs the target gas supply demand value for each of the subsequent x consecutive prediction units. The target gas supply demand value represents the total gas supply required for the corresponding prediction unit. After obtaining the target gas supply demand value, the causal convolutional network further outputs the flow allocation results corresponding to each module operating unit. The load control center decomposes the target gas supply demand value according to the flow allocation results of each module operating unit to obtain the initial target output flow value of each module operating unit in the predicted unit time. The sum of the initial target output flow values ​​of all module operating units is calculated. If the sum is less than the target gas supply demand value in the predicted unit time, the difference is allocated to the module operating units with a target status mark of 2, and the allocation order is sorted from largest to smallest according to the initial target output flow value in the predicted unit time. If the sum is greater than the target gas supply demand value in the predicted unit time, the excess is deducted from the module operating units with a target status mark of 2, and the deduction order is sorted from largest to smallest according to the initial target output flow value in the predicted unit time, until the sum of the target output flow values ​​of all module operating units equals the target gas supply demand value in the predicted unit time. After obtaining the initial target output flow rate value, the target status flag value is determined. For any module operating unit at any prediction unit time, if the initial target output flow rate value is equal to 0, then the module operating unit will not undertake the gas supply task at that prediction unit time; if the initial target output flow rate value is greater than 0, then the module operating unit will undertake the gas supply task at that prediction unit time. The load control center then combines the changes in the initial target output flow rate value before and after the prediction unit time to determine the target status flag value as 0, 1 or 2. Obtain the operating frequency correspondence table, which is a table formed by the collection of operating frequencies corresponding to different output flow values ​​of the module operating units; the load control center converts the target output flow value of each module operating unit into the target operating frequency value based on the output flow and operating frequency correspondence table of each module operating unit.

[0010] Preferably, when the target status flag value of a module operating unit changes from 0 to 1 or 2 in two consecutive predicted time units, the next predicted time unit is determined as the start-up time point; when it changes from 1 or 2 to 0, the next predicted time unit is determined as the stop-up time point; when it changes from 1 to 2, the next predicted time unit is determined as the loading switching time point; when it changes from 2 to 1, the next predicted time unit is determined as the unloading switching time point. The order of deployment is obtained according to the order in which each module operating unit first appears at its start-up time point. If the first start-up time points are the same, they are sorted from largest to smallest according to the target output flow value at that time point. The order of exit is obtained according to the order in which each module operating unit first appears at its stop-up time point. If the first stop-up time points are the same, they are sorted from smallest to largest according to the target output flow value in the predicted time unit before the stop-up. The load control center generates multiple initial candidate control solutions based on the target gas supply demand value, the target output flow value of each module operating unit, the target operating frequency value, the start-up time, the shutdown time, the load switching time, the unloading switching time, the input sequence, and the exit sequence, and sends them to S3.

[0011] Preferably, the load control center uses the multiple initial candidate control solutions obtained in S2 as optimization objects, and for each initial candidate control solution, calculates the cumulative value of gas supply deviation, the cumulative number of switching times, and the cumulative value of frequency over the subsequent x prediction unit times: For each prediction unit time, first calculate the sum of the target output flow values ​​of all module operating units, then subtract it from the target gas supply demand value corresponding to that prediction unit time, and take the absolute value of the difference; accumulate the absolute values ​​corresponding to x consecutive prediction units time to obtain the cumulative gas supply deviation value of the initial candidate control solution; The control input data of the module operating unit corresponding to the current unit time is compared with the target state marker value corresponding to the first prediction unit time. Then, the target state marker values ​​corresponding to the two prediction units before and after are compared one by one. Each time the target state marker value changes, it is recorded as one switch. The number of switches of all module operating units at the above comparison positions is accumulated to obtain the cumulative number of switches of the initial candidate control solution. The target operating frequency values ​​of all module operating units are accumulated one by one over x consecutive prediction unit times to obtain the cumulative frequency value of the initial candidate control solution. Sort the initial candidate control solutions by cumulative gas supply deviation value from smallest to largest, and retain the one with the smallest cumulative gas supply deviation value. If the smallest value corresponds to more than two initial candidate control solutions, sort them by cumulative switching count from smallest to largest, and retain the one with the smallest cumulative switching count. If the cumulative switching count is still the same, sort them by cumulative frequency value from smallest to largest, and retain the one with the smallest cumulative frequency value. When at least two initial candidate control solutions are retained, the load control center generates new candidate control solutions. These solutions are processed sequentially for x consecutive prediction units. For each prediction unit, the target state flag value from the initial candidate control solution with the smaller cumulative number of switching operations is taken as the target state flag value for that prediction unit. Then, the target output flow rate value from the initial candidate control solution with the smaller cumulative gas supply deviation value is taken as the initial target output flow rate value for that prediction unit. If the selected target state flag value is 0, the target output flow rate value for the corresponding module operating unit is set to 0. If the selected target state flag value is 1 or 2, the corresponding target output flow rate value is retained. Afterwards, the system is recalibrated according to the rule that the sum of the target output flow rates of all module operating units equals the target gas supply demand value, and the target operating frequency value is recalculated, thus obtaining a new candidate control solution.

[0012] Preferably, after obtaining a new candidate control solution, the load control center further corrects it: For any module operating unit with a prediction unit time, the module operating units are sorted from smallest to largest according to the target output flow value corresponding to that prediction unit time. Two module operating units with a target state flag value of 2 are selected sequentially. The target output flow value of one module operating unit is increased by one flow step, while the target output flow value of the other module operating unit is decreased by one flow step, with the increase and decrease values ​​being equal. The flow step is the difference between the output flow value of the corresponding module operating unit and the output flow value of two adjacent levels in the pre-stored output flow and operating frequency correspondence table. After each correction, the target operating frequency value is recalculated. If any of the recalculated target operating frequency values ​​exceeds the allowable output range of the corresponding strain gauge, the correction is cancelled. Multiple corrected candidate control solutions are generated in this way. The load control center recalculates the cumulative gas supply deviation, cumulative number of switching operations, and cumulative frequency for the retained initial candidate control solutions, new candidate control solutions, and modified candidate control solutions. The final result is determined in the order of minimum cumulative gas supply deviation, minimum cumulative number of switching operations, and minimum cumulative frequency. The determined result is used as the target candidate control solution and sent to S4.

[0013] Preferably, the load control center converts the target candidate control solution determined in S3 into control commands and sends them to each module operation unit for execution through the Internet of Things communication layer. The control commands include at least: start command, stop command, operating frequency setting command, loading command, and unloading command for the corresponding module operation unit. Each module operation unit performs the corresponding operation according to the start time, stop time, loading switching time, unloading switching time, input order, and exit order in the target candidate control solution. After execution, the load control center continues to acquire outlet pressure data, output flow data, operating frequency data, loading status data, unloading status data, and start / stop status data of each module's operating unit through the IoT communication layer, as well as the total gas consumption data and pipeline pressure data of the main pipeline. The load control center compares the acquired data with the target output flow value, target operating frequency value, and target status flag value in the target candidate control solution. If the actual execution result meets the current control requirements, the actual execution result is used as one of the input data for the next cycle. If the actual execution result is inconsistent with the target candidate control solution, S1 to S4 are re-executed in the next cycle.

[0014] The host modular air compressor load intelligent control system based on IoT communication includes multiple module operating units, an IoT communication layer, and a load control center: The modular operating unit is a functional module in the main modular air compressor system that can independently start, stop, load, unload, and output compressed air; The Internet of Things (IoT) communication layer is used to realize the transmission of operational data and control commands between the operating units of each module and the load control center. The load control center has a built-in load control model based on causal convolutional networks. This model receives the operating data of the current unit time and the preceding consecutive unit time, extracts the temporal correlation features, and outputs the target gas supply demand and control suggestions for subsequent consecutive unit time periods.

[0015] (III) Beneficial Effects This invention provides a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication, which has the following advantages: 1. By using a causal convolutional network, the total gas flow rate data, pipeline pressure data, and output flow rate data of each module operation unit are extracted from the current unit time and the previous multiple consecutive unit time data. Based on this, the target gas supply demand value for subsequent consecutive unit time is generated. This makes the judgment of gas supply demand no longer dependent on data at a single moment, which is conducive to improving the ability to identify gas consumption change trends and the foresight of subsequent regulation.

[0016] 2. The initial target output flow rate of each module operating unit is output using a causal convolutional network. Then, the load control center corrects the output flow rate according to the target gas supply demand value, so that the sum of the target output flow rate of each module operating unit is consistent with the target gas supply demand value. This enables coordinated allocation among multiple module operating units while meeting the total gas supply demand, reducing insufficient or redundant gas supply and improving the rationality of load allocation.

[0017] 3. By calculating the total gas supply deviation, the total number of state transitions, and the total operating frequency of the initial candidate control solutions, and then screening them in the order of priority of total gas supply deviation, followed by total number of state transitions, and then total operating frequency, it is possible to reduce unnecessary start-up, shutdown, loading and unloading transitions while meeting the target gas supply demand, which is conducive to improving the stability of system operation.

[0018] 4. By optimizing multiple initial candidate control solutions and determining the target candidate control solution, the final execution scheme can simultaneously take into account the degree of satisfaction of the target air supply demand, the effect of state switching control, and the effect of operating frequency control. This can reduce the overall consumption of the main modular air compressor system during operation, improve the collaborative operating efficiency of each module unit, and enhance the intelligent control level of the entire system. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the intelligent load control method for a modular air compressor based on Internet of Things communication according to the present invention. Figure 2 This is a schematic diagram of the host modular air compressor load intelligent control system based on Internet of Things communication according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 2This invention provides a modular air compressor load intelligent control system based on Internet of Things (IoT) communication, comprising multiple module operating units, an IoT communication layer, and a load control center. The modular operating unit is a functional module in the main modular air compressor system that can independently start, stop, load, unload, and output compressed air; The Internet of Things (IoT) communication layer is used to realize the transmission of operational data and control commands between the operating units of each module and the load control center. The load control center has a built-in load control model based on causal convolutional networks. This model receives the operating data of the current unit time and the preceding consecutive unit time, extracts the temporal correlation features, and outputs the target gas supply demand and control suggestions for subsequent consecutive unit time periods.

[0022] Please see Figure 1 This invention provides a method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication, comprising the following steps: S1. Collect operational data and generate control input data.

[0023] The system periodically acquires the current unit-time control input data and multiple historical control input data of the modular air compressor system via the IoT communication layer and sends them to the load control center. The control input data and multiple historical control input data include: output flow data, operating frequency data, status flag data, loading status data, unloading status data, and start / stop status data periodically collected by each module operating unit, as well as the total air consumption data of the main pipeline. Among them, the output flow data is acquired by the flow sensor installed on the outlet pipeline of the corresponding module operating unit, the operating frequency data is acquired by the frequency converter feedback signal of the corresponding module operating unit, the loading status data, unloading status data, and start / stop status data are acquired by the controller status feedback signal of the corresponding module operating unit, and the total air consumption data is acquired by the flow sensor installed on the main pipeline.

[0024] S2. Generate initial candidate control solutions based on the control input data through a causal convolutional network.

[0025] The load control center inputs control input data and multiple historical control input data into a causal convolutional network. This causal convolutional network is capable of extracting temporal correlation features and makes predictions for x subsequent time units as prediction units. At each prediction unit, the causal convolutional network outputs the target gas supply demand value, the initial target output flow rate value of each module operating unit, the target status marker value, the start-up time, the shutdown time, the loading switch time, the unloading switch time, the input sequence, and the exit sequence. The target status marker value is a discrete value representing the target operating state of each module operating unit at each prediction unit. For any module operating unit at any prediction unit, the target status marker value is 0, 1, or 2. In the target status marker value, 0 indicates shutdown, 1 indicates unloading operation, and 2 indicates loading operation. Where x is a positive integer greater than 1, which can be adjusted according to the actual situation; The causal convolutional network receives regulatory input data and multiple historical regulatory input data. It performs convolution operations on the regulatory input data and multiple historical regulatory input data in chronological order to extract the temporal relationship between gas consumption data and output flow data. Based on this temporal relationship, the causal convolutional network outputs the target gas supply demand value for each of the subsequent x consecutive prediction units. The target gas supply demand value represents the total gas supply required for the corresponding prediction unit. After obtaining the target gas supply demand value, the causal convolutional network further outputs the flow allocation results corresponding to each module operating unit. The load control center decomposes the target gas supply demand value according to the flow allocation results of each module operating unit to obtain the initial target output flow value of each module operating unit in the predicted unit time. The sum of the initial target output flow values ​​of all module operating units is calculated. If the sum is less than the target gas supply demand value in the predicted unit time, the difference is allocated to the module operating units with a target status mark of 2, and the allocation order is sorted from largest to smallest according to the initial target output flow value in the predicted unit time. If the sum is greater than the target gas supply demand value in the predicted unit time, the excess is deducted from the module operating units with a target status mark of 2, and the deduction order is sorted from largest to smallest according to the initial target output flow value in the predicted unit time, until the sum of the target output flow values ​​of all module operating units equals the target gas supply demand value in the predicted unit time. After obtaining the initial target output flow rate value, the target status flag value is determined. For any module operating unit at any prediction unit time, if the initial target output flow rate value is equal to 0, then the module operating unit will not undertake the gas supply task at that prediction unit time; if the initial target output flow rate value is greater than 0, then the module operating unit will undertake the gas supply task at that prediction unit time. The load control center then combines the changes in the initial target output flow rate value before and after the prediction unit time to determine the target status flag value as 0, 1 or 2. Obtain the operating frequency correspondence table, which is a table formed by the collection of operating frequencies corresponding to different output flow values ​​of the module operating units; the load control center converts the target output flow value of each module operating unit into the target operating frequency value based on the output flow and operating frequency correspondence table of each module operating unit.

[0026] When the target status flag value of a module's operating unit changes from 0 to 1 or 2 within two consecutive predicted time units, the subsequent predicted time unit is determined as the start-up time point; when it changes from 1 or 2 to 0, the subsequent predicted time unit is determined as the stop-up time point; when it changes from 1 to 2, the subsequent predicted time unit is determined as the loading switch-up time point; when it changes from 2 to 1, the subsequent predicted time unit is determined as the unloading switch-up time point. The order of deployment is determined by the order in which each module's operating unit first appears at its start-up time point. If the first start-up time points are the same, they are sorted by the target output flow value at that time point from largest to smallest. The order of exit is determined by the order in which each module's operating unit first appears at its stop-up time point. If the first stop-up time points are the same, they are sorted by the target output flow value at the predicted time unit before the stop-up time point from smallest to largest. The load control center generates multiple initial candidate control solutions based on the target gas supply demand value, the target output flow value of each module operating unit, the target operating frequency value, the start-up time, the shutdown time, the load switching time, the unloading switching time, the input sequence, and the exit sequence, and sends them to S3.

[0027] S3. Optimize the initial candidate control solution and determine the target candidate control solution.

[0028] The load control center uses the multiple initial candidate control solutions obtained from S2 as optimization objects. For each initial candidate control solution, it calculates the cumulative gas supply deviation, cumulative number of switching operations, and cumulative frequency over the next x prediction unit times. For each prediction unit time, first calculate the sum of the target output flow values ​​of all module operating units, then subtract it from the target gas supply demand value corresponding to that prediction unit time, and take the absolute value of the difference; accumulate the absolute values ​​corresponding to x consecutive prediction units time to obtain the cumulative gas supply deviation value of the initial candidate control solution; The control input data of the module operating unit corresponding to the current unit time is compared with the target state marker value corresponding to the first prediction unit time. Then, the target state marker values ​​corresponding to the two prediction units before and after are compared one by one. Each time the target state marker value changes, it is recorded as one switch. The number of switches of all module operating units at the above comparison positions is accumulated to obtain the cumulative number of switches of the initial candidate control solution. The target operating frequency values ​​of all module operating units are accumulated one by one over x consecutive prediction unit times to obtain the cumulative frequency value of the initial candidate control solution. Sort the initial candidate control solutions by cumulative gas supply deviation value from smallest to largest, and retain the one with the smallest cumulative gas supply deviation value. If the smallest value corresponds to more than two initial candidate control solutions, sort them by cumulative switching count from smallest to largest, and retain the one with the smallest cumulative switching count. If the cumulative switching count is still the same, sort them by cumulative frequency value from smallest to largest, and retain the one with the smallest cumulative frequency value. When at least two initial candidate control solutions remain, the load control center generates new candidate control solutions. These solutions are processed sequentially for x consecutive prediction units. For each prediction unit, the target state flag value from the initial candidate control solution with the smaller cumulative switching count is taken as the target state flag value for that prediction unit. Then, the target output flow rate value from the initial candidate control solution with the smaller cumulative gas supply deviation value is taken as the initial target output flow rate value for that prediction unit. If the selected target state flag value is 0, the target output flow rate value for the corresponding module operating unit is set to 0. If the selected target state flag value is 1 or 2, the corresponding target output flow rate value is retained. Afterwards, the system is recalibrated according to the rule that the sum of the target output flow rates of all module operating units equals the target gas supply demand value, and the target operating frequency value is recalculated, thus obtaining a new candidate control solution. After obtaining a new candidate control solution, the load control center further corrects it: For any module operating unit with a predicted unit time, the module operating units are sorted from smallest to largest according to the target output flow value corresponding to that predicted unit time. Two module operating units with a target status flag value of 2 are selected in sequence. The target output flow value of one module operating unit is increased by 1 flow step, while the target output flow value of the other module operating unit is decreased by 1 flow step, and the increase and decrease values ​​are equal. The flow step is the difference between the output flow value of the corresponding module operating unit and the output flow value of two adjacent levels in the table corresponding to the operating frequency. After each correction, the target operating frequency value is recalculated. The load control center recalculates the cumulative gas supply deviation, cumulative number of switching operations, and cumulative frequency for the retained initial candidate control solutions, new candidate control solutions, and modified candidate control solutions. The final result is determined in the order of minimum cumulative gas supply deviation, minimum cumulative number of switching operations, and minimum cumulative frequency. The determined final result is used as the target candidate control solution and sent to S4.

[0029] S4. Execute the target candidate control solution and perform rolling updates.

[0030] The load control center converts the target candidate control solution determined by S3 into control commands and sends them to each module operation unit for execution through the Internet of Things communication layer. The control commands include at least the start command, stop command, operating frequency setting command, loading command, and unloading command of the corresponding module operation unit. Each module operation unit performs the corresponding operation according to the start time, stop time, loading switching time, unloading switching time, input order, and exit order in the target candidate control solution. After execution, the load control center continues to acquire outlet pressure data, output flow data, operating frequency data, loading status data, unloading status data, and start / stop status data of each module's operating unit through the IoT communication layer, as well as the total gas consumption data and pipeline pressure data of the main pipeline. The load control center compares the acquired data with the target output flow value, target operating frequency value, and target status flag value in the target candidate control solution. If the actual execution result meets the current control requirements, the actual execution result is used as one of the input data for the next cycle. If the actual execution result is inconsistent with the target candidate control solution, S1 to S4 are re-executed in the next cycle.

[0031] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0032] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0033] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent load control of a modular air compressor based on Internet of Things (IoT) communication, characterized in that: include: S1. Collect operational data and generate control input data; S2. Generate initial candidate control solutions based on the control input data using a causal convolutional network; S3. Optimize the initial candidate control solution and determine the target candidate control solution; S4. Execute the target candidate control solution and perform rolling updates.

2. The method for intelligent load control of a modular air compressor based on Internet of Things communication according to claim 1, characterized in that: In S1, operational data is collected and used to generate control input data, including: The system periodically acquires the current unit-time control input data and multiple historical control input data of the modular air compressor system via the IoT communication layer and sends them to the load control center. The control input data and multiple historical control input data include: output flow data, operating frequency data, status flag data, loading status data, unloading status data, and start / stop status data periodically collected by each module operating unit, as well as the total air consumption data of the main pipeline. Among them, the output flow data is acquired by the flow sensor installed on the outlet pipeline of the corresponding module operating unit, the operating frequency data is acquired by the frequency converter feedback signal of the corresponding module operating unit, the loading status data, unloading status data, and start / stop status data are acquired by the controller status feedback signal of the corresponding module operating unit, and the total air consumption data is acquired by the flow sensor installed on the main pipeline.

3. The method for intelligent load control of a modular air compressor based on Internet of Things communication according to claim 1, characterized in that: In S2, initial candidate control solutions are generated through a causal convolutional network, including: The load control center inputs control input data and multiple historical control input data into a causal convolutional network. This causal convolutional network is capable of extracting temporal correlation features and makes predictions for x subsequent time units as prediction units. At each prediction unit, the causal convolutional network outputs the target gas supply demand value, the initial target output flow rate value of each module operating unit, the target status marker value, the start-up time, the shutdown time, the loading switch time, the unloading switch time, the input sequence, and the exit sequence. The target status marker value is a discrete value representing the target operating state of each module operating unit at each prediction unit. For any module operating unit at any prediction unit, the target status marker value is 0, 1, or 2. In the target status marker value, 0 indicates shutdown, 1 indicates unloading operation, and 2 indicates loading operation. Where x is a positive integer greater than 1, which can be adjusted according to the actual situation; The causal convolutional network receives regulatory input data and multiple historical regulatory input data. It performs convolution operations on the regulatory input data and multiple historical regulatory input data in chronological order to extract the temporal relationship between gas consumption data and output flow data. Based on this temporal relationship, the causal convolutional network outputs the target gas supply demand value for each of the subsequent x consecutive prediction units. The target gas supply demand value represents the total gas supply required for the corresponding prediction unit. After obtaining the target gas supply demand value, the causal convolutional network further outputs the flow allocation results corresponding to each module operating unit. The load control center decomposes the target gas supply demand value according to the flow allocation results of each module operating unit to obtain the initial target output flow value of each module operating unit in the predicted unit time. The sum of the initial target output flow values ​​of all module operating units is calculated. If the sum is less than the target gas supply demand value in the predicted unit time, the difference is allocated to the module operating units with a target status mark of 2, and the allocation order is sorted from largest to smallest according to the initial target output flow value in the predicted unit time. If the sum is greater than the target gas supply demand value in the predicted unit time, the excess is deducted from the module operating units with a target status mark of 2, and the deduction order is sorted from largest to smallest according to the initial target output flow value in the predicted unit time, until the sum of the target output flow values ​​of all module operating units equals the target gas supply demand value in the predicted unit time. After obtaining the initial target output flow rate value, the target status flag value is determined. For any module operating unit at any prediction unit time, if the initial target output flow rate value is equal to 0, then the module operating unit will not undertake the gas supply task at that prediction unit time; if the initial target output flow rate value is greater than 0, then the module operating unit will undertake the gas supply task at that prediction unit time. The load control center then combines the changes in the initial target output flow rate value before and after the prediction unit time to determine the target status flag value as 0, 1 or 2. Obtain the operating frequency correspondence table, which is a table formed by the collection of operating frequencies corresponding to different output flow values ​​of the module operating units; the load control center converts the target output flow value of each module operating unit into the target operating frequency value based on the output flow and operating frequency correspondence table of each module operating unit.

4. The method for intelligent load control of a modular air compressor based on Internet of Things communication according to claim 1, characterized in that: In S2, initial candidate control solutions are generated through a causal convolutional network, including: When the target status flag value of a module's operating unit changes from 0 to 1 or 2 within two consecutive predicted time units, the subsequent predicted time unit is determined as the start-up time point; when it changes from 1 or 2 to 0, the subsequent predicted time unit is determined as the stop-up time point; when it changes from 1 to 2, the subsequent predicted time unit is determined as the loading switch-up time point; when it changes from 2 to 1, the subsequent predicted time unit is determined as the unloading switch-up time point. The order of deployment is determined by the order in which each module's operating unit first appears at its start-up time point. If the first start-up time points are the same, they are sorted by the target output flow value at that time point from largest to smallest. The order of exit is determined by the order in which each module's operating unit first appears at its stop-up time point. If the first stop-up time points are the same, they are sorted by the target output flow value at the predicted time unit before the stop-up time point from smallest to largest. The load control center generates multiple initial candidate control solutions based on the target gas supply demand value, the target output flow value of each module operating unit, the target operating frequency value, the start-up time, the shutdown time, the load switching time, the unloading switching time, the input sequence, and the exit sequence, and sends them to S3.

5. The method for intelligent load control of a modular air compressor based on Internet of Things communication according to claim 1, characterized in that: In S3, the initial candidate control solution is optimized and the target candidate control solution is determined, including: The load control center uses the multiple initial candidate control solutions obtained from S2 as optimization objects. For each initial candidate control solution, it calculates the cumulative gas supply deviation, cumulative number of switching operations, and cumulative frequency over the next x prediction unit times. For each prediction unit time, first calculate the sum of the target output flow values ​​of all module operating units, then subtract it from the target gas supply demand value corresponding to that prediction unit time, and take the absolute value of the difference; accumulate the absolute values ​​corresponding to x consecutive prediction units time to obtain the cumulative gas supply deviation value of the initial candidate control solution; The control input data of the module operating unit corresponding to the current unit time is compared with the target state marker value corresponding to the first prediction unit time. Then, the target state marker values ​​corresponding to the two prediction units before and after are compared one by one. Each time the target state marker value changes, it is recorded as one switch. The number of switches of all module operating units at the above comparison positions is accumulated to obtain the cumulative number of switches of the initial candidate control solution. The target operating frequency values ​​of all module operating units are accumulated one by one over x consecutive prediction unit times to obtain the cumulative frequency value of the initial candidate control solution. Sort the initial candidate control solutions by cumulative gas supply deviation value from smallest to largest, and retain the one with the smallest cumulative gas supply deviation value. If the smallest value corresponds to more than two initial candidate control solutions, sort them by cumulative switching count from smallest to largest, and retain the one with the smallest cumulative switching count. If the cumulative switching count is still the same, sort them by cumulative frequency value from smallest to largest, and retain the one with the smallest cumulative frequency value. When at least two initial candidate control solutions are retained, the load control center generates new candidate control solutions. These solutions are processed sequentially for x consecutive prediction units. For each prediction unit, the target state flag value from the initial candidate control solution with the smaller cumulative number of switching operations is taken as the target state flag value for that prediction unit. Then, the target output flow rate value from the initial candidate control solution with the smaller cumulative gas supply deviation value is taken as the initial target output flow rate value for that prediction unit. If the selected target state flag value is 0, the target output flow rate value for the corresponding module operating unit is set to 0. If the selected target state flag value is 1 or 2, the corresponding target output flow rate value is retained. Afterwards, the system is recalibrated according to the rule that the sum of the target output flow rates of all module operating units equals the target gas supply demand value, and the target operating frequency value is recalculated, thus obtaining a new candidate control solution.

6. The method for intelligent load control of a modular air compressor based on Internet of Things communication according to claim 1, characterized in that: In S3, the initial candidate control solution is optimized and the target candidate control solution is determined, including: After obtaining a new candidate control solution, the load control center further refines it: For any module operating unit with a predicted unit time, the module operating units are sorted from smallest to largest according to the target output flow value corresponding to that predicted unit time. Two module operating units with a target state flag value of 2 are selected sequentially. The target output flow value of one module operating unit is increased by one flow step, while the target output flow value of the other module operating unit is decreased by one flow step, with the increase and decrease values ​​being equal. The flow step is the difference between the output flow value of the corresponding module operating unit and the output flow value of two adjacent levels in the pre-stored output flow and operating frequency correspondence table. After each refinement, the target operating frequency value is recalculated. If any of the recalculated target operating frequency values ​​exceeds the allowable output range of the corresponding strain gauge, the refinement is cancelled. Multiple refined candidate control solutions are generated through this method. The load control center recalculates the cumulative gas supply deviation, cumulative number of switching operations, and cumulative frequency for the retained initial candidate control solutions, new candidate control solutions, and modified candidate control solutions. The final result is determined in the order of minimum cumulative gas supply deviation, minimum cumulative number of switching operations, and minimum cumulative frequency. The determined result is used as the target candidate control solution and sent to S4.

7. The method for intelligent load control of a modular air compressor based on Internet of Things communication according to claim 1, characterized in that: In S4, the target candidate control solution is executed and rolled updates are performed, including: The load control center converts the target candidate control solution determined by S3 into control commands and sends them to each module operation unit for execution through the Internet of Things communication layer. The control commands include at least the start command, stop command, operating frequency setting command, loading command, and unloading command of the corresponding module operation unit. Each module operation unit performs the corresponding operation according to the start time, stop time, loading switching time, unloading switching time, input order, and exit order in the target candidate control solution. After execution, the load control center continues to acquire outlet pressure data, output flow data, operating frequency data, loading status data, unloading status data, and start / stop status data of each module's operating unit through the IoT communication layer, as well as the total gas consumption data and pipeline pressure data of the main pipeline. The load control center compares the acquired data with the target output flow value, target operating frequency value, and target status flag value in the target candidate control solution. If the actual execution result meets the current control requirements, the actual execution result is used as one of the input data for the next cycle. If the actual execution result is inconsistent with the target candidate control solution, S1 to S4 are re-executed in the next cycle.

8. The host modular air compressor load intelligent control system based on Internet of Things communication according to claim 1, characterized in that, It includes multiple module operation units, an IoT communication layer, and a load control center: The modular operating unit is a functional module in the main modular air compressor system that can independently start, stop, load, unload, and output compressed air; The Internet of Things (IoT) communication layer is used to realize the transmission of operational data and control commands between the operating units of each module and the load control center. The load control center has a built-in load control model based on causal convolutional networks. This model receives the operating data of the current unit time and the preceding consecutive unit time, extracts the temporal correlation features, and outputs the target gas supply demand and control suggestions for subsequent consecutive unit time periods.

Images