Air compressor scheduling optimization method and system based on artificial intelligence and cloud edge collaboration

Abstract

The application provides an air compressor optimization method and system based on artificial intelligence and cloud edge collaboration. The method comprises the following steps: using a preset full connection neural network algorithm to model an air compressor performance model, obtaining a power-yield neural network model and a power-frequency neural network model, and storing them in the cloud; the edge end adjusts the frequency of the air compressor according to the model; using a preset long short-term memory artificial neural network algorithm to construct a demand prediction model and storing it in the cloud; the edge end predicts the current demand of the air compressor unit according to the demand prediction model, obtains the expected demand, calculates the expected total power of the air compressor unit when the air compressor unit reaches the expected demand, obtains the full load flow rate of the air compressor, compares the expected demand with the full load flow rate, selects the best scheduling optimization scheme of the air compressor unit according to the comparison result, and realizes that the air compressor unit is always in the optimal efficiency state under the working condition, and achieves the purpose of energy saving and cost reduction.

Description

Technical Field

[0001] This application relates to the field of intelligent optimization and adjustment technology for air compressor units, specifically, to an air compressor adjustment and optimization method and system based on artificial intelligence and cloud-edge collaboration. Background Technology

[0002] Traditional variable frequency air compressors typically use a proportional-integral-derivative (PID) algorithm to control the frequency conversion based on the set exhaust pressure and the current pressure. In actual engineering projects, multiple air compressors usually operate in conjunction. Due to differences in pipeline and air tank connection locations, sensor errors, and other issues, some air compressors with the same PID parameters may be more sensitive, leading to high loads on sensitive compressors and low loads on sluggish compressors. This results in the system not operating in an optimal state. Furthermore, different operating combinations will have different sensitivities, making optimization impossible by adjusting PID parameters individually. Existing unit scheduling methods generally employ full-load methods (adding a machine when all operating air compressors reach their rated power or frequency, automatically reducing the frequency of air compressors until unloading when supply exceeds demand) or pressure control (starting one compressor when the mains pressure is below a certain threshold or shutting down one compressor when it is above a certain threshold). Such scheduling methods are somewhat arbitrary, failing to consider the different actual efficiencies of multiple air compressors, failing to prioritize starting high-efficiency compressors, failing to prioritize stopping low-efficiency compressors, and failing to ensure the air compressor unit operates at its optimal efficiency.

[0003] Effective technical solutions are urgently needed to address the above problems. Summary of the Invention

[0004] This invention provides a method for frequency conversion regulation and start-stop scheduling of a variable frequency air compressor unit. The method includes using a fully connected neural network (FCNN) in the cloud to model the air compressor performance model, using a long short-term memory neural network (LSTM) to model the compressed air demand, and performing frequency conversion optimization and scheduling optimization at the edge based on the performance model and demand model synchronized in the cloud, so as to ensure that the air compressor unit is always in the optimal efficiency state under the operating conditions, thereby achieving the purpose of energy saving and cost reduction.

[0005] This application also provides a method for optimizing air compressors based on artificial intelligence and cloud-edge collaboration, including the following steps:

[0006] The performance model of the air compressor is modeled using a pre-defined fully connected neural network algorithm, resulting in a power-output neural network model and a power-frequency neural network model, which are then stored in the cloud.

[0007] The edge device adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model.

[0008] A demand prediction model is constructed using a pre-defined long short-term memory artificial neural network algorithm and stored in the cloud. The edge device predicts the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand.

[0009] The edge calculation determines the expected total power of the air compressor unit when it reaches the expected demand.

[0010] The full-load flow rate of the air compressor is obtained at the edge, the expected demand is compared with the full-load flow rate, and the best scheduling optimization scheme of the air compressor unit is selected based on the comparison results.

[0011] Optionally, in the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration described in this application, the step of using a preset fully connected neural network algorithm to model the air compressor performance model, obtaining a power-output neural network model and a power-frequency neural network model, and storing them in the cloud includes:

[0012] Real-time acquisition of air compressor operating data, including instantaneous power, instantaneous frequency, status marker data, and average flow rate;

[0013] Based on a preset fully connected neural network algorithm, the instantaneous power, state marker data and average flow rate of a single air compressor are used to initialize the parameters of the preset power-output model, resulting in an initialized power-output neural network model.

[0014] The total average flow velocity of the air compressor unit is calculated based on the average flow velocity of all air compressors.

[0015] The initial power-output neural network model is trained based on the total average flow rate and the instantaneous power and state label data of each air compressor to obtain the power-output neural network model.

[0016] A power-frequency neural network model is obtained by training the model using a preset fully connected neural network algorithm and the instantaneous power and instantaneous frequency of each air compressor.

[0017] The power-output neural network model and the power-frequency neural network model are stored in the cloud.

[0018] Optionally, in the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration described in this application, the edge device adjusts the frequency of the air compressor according to the power-output neural network model and the power-frequency neural network model, including:

[0019] The edge terminal acquires the current pressure value, current power value, current flow rate, and current status flag data of the air compressor;

[0020] The system power increase or decrease is obtained by performing proportional-integral calculations based on the current pressure value and the set pressure value, and the adjustment power value and adjustment number set value are determined.

[0021] The regulated power value and the current state label data are input into the power-output neural network model for processing to obtain the regulated flow rate;

[0022] Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0023] The power adjustment value is input into the power-frequency neural network model to obtain the adjustment frequency value. The target air compressor frequency converter is set to the adjustment frequency value, and the current adjustment count is updated synchronously.

[0024] The current number of adjustments is compared with the set number of adjustments. If the current number of adjustments is less than the set number of adjustments, the adjustment continues to repeat.

[0025] Optionally, in the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration described in this application, the step of constructing a demand prediction model using a preset long short-term memory artificial neural network algorithm and storing it in the cloud, and then using the edge to predict the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand, includes:

[0026] Obtain the average demand of the air compressor and the average flow rate and instantaneous pressure within a preset time period;

[0027] Based on a preset long short-term memory artificial neural network algorithm, and based on the average flow rate, instantaneous pressure and average demand within a preset time period, the model is trained to obtain a demand prediction model and store it in the cloud.

[0028] The average flow rate and instantaneous pressure of the air compressor unit within the current preset time period are obtained at the edge end;

[0029] The average flow rate and instantaneous pressure within the current preset time period are input into the preset demand prediction model to obtain the expected demand.

[0030] Optionally, in the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration described in this application, the edge computing of the expected total power of the air compressor unit when it reaches the expected demand includes:

[0031] The efficiency of each air compressor at different power levels is obtained at the edge, and the power value with the highest efficiency and the average flow rate are selected as the optimal power value and the optimal output value.

[0032] The optimal total power of the air compressor unit is obtained by summing the optimal power values ​​of all air compressors, and the optimal total output value of the air compressor unit is obtained by summing the optimal output values ​​of all air compressors.

[0033] The adjustment power value is determined based on the comparison between the expected demand and the optimal total output;

[0034] The adjusted power value and current state label data are input into the power-output neural network model for processing to obtain the adjusted flow rate;

[0035] Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0036] The adjusted power value is input into the power-frequency neural network model for processing to obtain the adjusted frequency value. The target air compressor inverter is set to the adjusted frequency value, and the average flow rate of the adjusted air compressor unit is calculated based on the adjusted flow rate of the target air compressor.

[0037] Compare the adjusted average flow rate of the air compressor unit with the expected demand. If the comparison is consistent, the total power of the air compressor unit at this time is taken as the expected total power of the air compressor unit.

[0038] If the comparison is inconsistent, continue to adjust until the average flow rate of the adjusted air compressor unit reaches the expected demand. Then, take the total power of the air compressor unit at this time as the expected total power of the air compressor unit.

[0039] Optionally, in the air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration described in this application, the edge device obtains the full-load flow rate of the air compressor, compares the expected demand with the full-load flow rate, and selects the optimal scheduling optimization scheme for the air compressor unit based on the comparison result, including:

[0040] If the full-load flow rate is greater than the expected demand, calculate the expected total power of the air compressor unit with one additional air compressor and two additional air compressors respectively, and select the scheme with the lowest expected total power to be implemented.

[0041] If the full-load flow rate is less than the expected demand, calculate the expected total power and energy saving rate of the air compressor unit by adding one, removing one, and replacing one air compressor respectively, and select the scheme with the highest energy saving rate of greater than 5% to implement.

[0042] Secondly, this application provides an air compressor tuning and optimization system based on artificial intelligence and cloud-edge collaboration. The system includes a memory and a processor. The memory stores a program for an air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration. When the program for the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration is executed by the processor, it performs the following steps:

[0043] The performance model of the air compressor is modeled using a pre-defined fully connected neural network algorithm, resulting in a power-output neural network model and a power-frequency neural network model, which are then stored in the cloud.

[0044] The edge device adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model.

[0045] A demand prediction model is constructed using a pre-defined long short-term memory artificial neural network algorithm and stored in the cloud. The edge device predicts the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand.

[0046] The edge calculation determines the expected total power of the air compressor unit when it reaches the expected demand.

[0047] The full-load flow rate of the air compressor is obtained at the edge, the expected demand is compared with the full-load flow rate, and the best scheduling optimization scheme of the air compressor unit is selected based on the comparison results.

[0048] Optionally, in the air compressor tuning and optimization system based on artificial intelligence and cloud-edge collaboration described in this application, the step of using a preset fully connected neural network algorithm to model the air compressor performance model, obtaining a power-output neural network model and a power-frequency neural network model, and storing them in the cloud includes:

[0049] Real-time acquisition of air compressor operating data, including instantaneous power, instantaneous frequency, status marker data, and average flow rate;

[0050] Based on a preset fully connected neural network algorithm, the instantaneous power, state marker data and average flow rate of a single air compressor are used to initialize the parameters of the preset power-output model, resulting in an initialized power-output neural network model.

[0051] The total average flow velocity of the air compressor unit is calculated based on the average flow velocity of all air compressors.

[0052] The initial power-output neural network model is trained based on the total average flow rate and the instantaneous power and state label data of each air compressor to obtain the power-output neural network model.

[0053] A power-frequency neural network model is obtained by training the model using a preset fully connected neural network algorithm and the instantaneous power and instantaneous frequency of each air compressor.

[0054] The power-output neural network model and the power-frequency neural network model are stored in the cloud.

[0055] Optionally, in the air compressor tuning and optimization system based on artificial intelligence and cloud-edge collaboration described in this application, the edge device adjusts the frequency of the air compressor according to the power-output neural network model and the power-frequency neural network model, including:

[0056] The edge terminal acquires the current pressure value, current power value, current flow rate, and current status flag data of the air compressor;

[0057] The system power increase or decrease is obtained by performing proportional-integral calculations based on the current pressure value and the set pressure value, and the adjustment power value and adjustment number set value are determined.

[0058] The regulated power value and the current state label data are input into the power-output neural network model for processing to obtain the regulated flow rate;

[0059] Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0060] The power adjustment value is input into the power-frequency neural network model to obtain the adjustment frequency value. The target air compressor frequency converter is set to the adjustment frequency value, and the current adjustment count is updated synchronously.

[0061] The current number of adjustments is compared with the set number of adjustments. If the current number of adjustments is less than the set number of adjustments, the adjustment continues to repeat.

[0062] Optionally, in the air compressor adjustment and optimization system based on artificial intelligence and cloud-edge collaboration described in this application, the step of constructing a demand prediction model using a preset long short-term memory artificial neural network algorithm and storing it in the cloud, and then using the edge to predict the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand, includes:

[0063] Obtain the average demand of the air compressor and the average flow rate and instantaneous pressure within a preset time period;

[0064] Based on a preset long short-term memory artificial neural network algorithm, and based on the average flow rate, instantaneous pressure and average demand within a preset time period, the model is trained to obtain a demand prediction model and store it in the cloud.

[0065] The average flow rate and instantaneous pressure of the air compressor unit within the current preset time period are obtained at the edge end;

[0066] The average flow rate and instantaneous pressure within the current preset time period are input into the preset demand prediction model to obtain the expected demand.

[0067] As can be seen from the above, the air compressor tuning and optimization method and system based on artificial intelligence and cloud-edge collaboration provided in this application, by using a fully connected neural network (FCNN) to model the air compressor performance model in the cloud, and using a long short-term memory neural network (LSTM) to model the compressed air demand, and performing frequency conversion optimization and scheduling optimization at the edge based on the performance model and demand model synchronized in the cloud, ensures that the air compressor unit is always in the optimal efficiency state under the operating conditions, thereby achieving the purpose of energy saving and cost reduction.

[0068] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0069] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0070] Figure 1 A flowchart of an air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration provided in this application embodiment;

[0071] Figure 2 A flowchart illustrating the air compressor performance modeling process in the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration provided in this application embodiment;

[0072] Figure 3 A flowchart illustrating the frequency adjustment of an air compressor using an air compressor optimization method based on artificial intelligence and cloud-edge collaboration, provided in this application embodiment.

[0073] Figure 4 The flowchart illustrates the process of obtaining the expected demand using an air compressor optimization method based on artificial intelligence and cloud-edge collaboration, as provided in this application embodiment. Detailed Implementation

[0074] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0075] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0076] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration, as described in some embodiments of this application. This air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration is used in terminal devices, such as computers and mobile terminals. The air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration includes the following steps:

[0077] S11. Use a pre-defined fully connected neural network algorithm to model the performance of the air compressor, obtain the power-output neural network model and the power-frequency neural network model, and store them in the cloud.

[0078] S12. The edge end adjusts the frequency of the air compressor according to the power-output neural network model and the power-frequency neural network model.

[0079] S13. A demand prediction model is constructed using a preset long short-term memory artificial neural network algorithm and stored in the cloud. The edge device predicts the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand.

[0080] S14. Calculate the expected total power of the air compressor unit when the expected demand is reached at the edge end.

[0081] S15. Obtain the full-load flow rate of the air compressor at the edge, compare the expected demand with the full-load flow rate, and select the best scheduling optimization scheme for the air compressor unit based on the comparison results.

[0082] It should be noted that this application uses a fully connected neural network (FCNN) in the cloud to model the performance of the air compressor and a long short-term memory neural network (LSTM) to model the compressed air demand. At the edge, frequency conversion optimization and scheduling optimization are performed based on the performance model and demand model synchronized in the cloud, so as to ensure that the air compressor unit is always in the optimal efficiency state under the operating conditions, thereby achieving the purpose of energy saving and cost reduction.

[0083] Please refer to Figure 2 , Figure 2This is a flowchart illustrating the air compressor performance modeling process in some embodiments of this application, specifically the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration. According to embodiments of the present invention, the step of using a preset fully connected neural network algorithm to model the air compressor performance, obtaining a power-output neural network model and a power-frequency neural network model, and storing them in the cloud includes:

[0084] S21. Real-time acquisition of air compressor operating data, including instantaneous power, instantaneous frequency, status marker data, and average flow rate;

[0085] S22. Based on the preset fully connected neural network algorithm, the instantaneous power, state marker data and average flow rate of a single air compressor are used to initialize the parameters of the preset power-output model to obtain the initialized power-output neural network model.

[0086] S23. Calculate the total average flow velocity of the air compressor unit based on the average flow velocity of all air compressors.

[0087] S24. Train the initial power-output neural network model based on the total average flow rate and the instantaneous power and status label data of each air compressor to obtain the power-output neural network model.

[0088] S25. Based on the preset fully connected neural network algorithm and using the instantaneous power and instantaneous frequency of each air compressor, a power-frequency neural network model is obtained.

[0089] S26. Store the power-output neural network model and the power-frequency neural network model in the cloud.

[0090] It should be noted that the status flag data represents the start / stop and loading / unloading states as two binary features: 1 for running and loading, and 0 for stopping or unloading. The average flow rate is the average flow rate of the air compressor over 5 seconds, and the instantaneous power is input at 0.5kW intervals. First, the model parameters are initialized using data from a single running unit. Then, data from all air compressors in the same unit running simultaneously is used for model training to obtain a power-output neural network model. The total average flow rate of the air compressor unit is the sum of the average flow rates of all air compressors over 5 seconds.

[0091] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating the frequency adjustment of an air compressor based on an artificial intelligence and cloud-edge collaboration-based air compressor optimization method in some embodiments of this application. According to embodiments of the present invention, the edge device adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model, including:

[0092] S31. The edge end acquires the current pressure value, current power value, current flow rate, and current status flag data of the air compressor;

[0093] S32. Perform proportional-integral calculations based on the current pressure value and the set pressure value to obtain the increase or decrease in system power, and determine the set value for the adjustment power value and the number of adjustments;

[0094] S33. Input the regulated power value and the current state label data into the power-output neural network model for processing to obtain the regulated flow rate;

[0095] S34. Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0096] S35. Input the adjusted power value into the power-frequency neural network model to obtain the adjusted frequency value, set the target air compressor frequency converter to the adjusted frequency value, and update the current number of adjustments synchronously.

[0097] S36. Compare the current number of adjustments with the set number of adjustments. If the current number of adjustments is less than the set number of adjustments, continue to repeat the adjustment.

[0098] It should be noted that the formula for calculating proportional integrals is:

[0099] ;

[0100] ;

[0101] in, Let be the power increase or decrease at time t. This is the proportionality coefficient. The integral coefficient is... The set pressure value at time t Let be the current pressure value at time t. For error, The integral of the error;

[0102] If the calculated value is an increase in system power, the sum of the current power value and 0.5KW is used as the adjustment power value. If the calculated value is a decrease in system power, the difference between the current power value and 0.5KW is used as the adjustment power value, and the ratio of the increase / decrease in system power to 0.5KW is used as the adjustment number. The difference between the adjusted flow rate and the current flow rate is calculated. If the calculated value is an increase in system power, the air compressor with the largest absolute value of the difference is used as the target air compressor. If the calculated value is a decrease in system power, the air compressor with the smallest absolute value of the difference is used as the target air compressor. If the proportional-integral (PI) calculation requires an increase in power, then based on the current power, the air compressor with the largest increase in output when power is increased by 0.5kW is selected, and the frequency of this air compressor is increased to achieve the increased power. If the PI calculation requires a decrease in power, then based on the current power, the air compressor with the smallest decrease in output when power is decreased by 0.5kW is selected, and the frequency of this air compressor is decreased to achieve the decreased power. This process is repeated multiple times until the power increase or decrease reaches the given power increase or decrease value, resulting in the optimized frequency for each air compressor. Each adjustment increments the current adjustment count by 1.

[0103] Please refer to Figure 4 , Figure 4 This is a flowchart illustrating the process of obtaining expected demand in an air compressor optimization method based on artificial intelligence and cloud-edge collaboration, as described in some embodiments of this application. According to embodiments of the present invention, the step of constructing a demand prediction model using a preset long short-term memory artificial neural network algorithm and storing it in the cloud, and then using the edge device to predict the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand, includes:

[0104] S41. Obtain the average demand of the air compressor and the average flow rate and instantaneous pressure within a preset time period;

[0105] S42. Based on a preset long short-term memory artificial neural network algorithm, and according to the average flow rate, instantaneous pressure and average demand within a preset time period, the model is trained to obtain a demand prediction model and store it in the cloud.

[0106] S43. The edge end acquires the average flow rate and instantaneous pressure of the air compressor unit within the current preset time period;

[0107] S44. Input the average flow velocity and instantaneous pressure within the current preset time period into the preset demand prediction model to obtain the expected demand.

[0108] It should be noted that the average flow rate and instantaneous pressure within the preset time period of the air compressor refer to the average flow rate over 15 seconds and the instantaneous pressure over the previous 60 minutes before the demand forecast time. The average demand refers to the average flow rate demand over the next 5 minutes before the demand forecast time.

[0109] According to an embodiment of the present invention, the calculation of the expected total power of the air compressor unit when the expected demand is reached at the edge end includes:

[0110] The efficiency of each air compressor at different power levels is obtained at the edge, and the power value with the highest efficiency and the average flow rate are selected as the optimal power value and the optimal output value.

[0111] The optimal total power of the air compressor unit is obtained by summing the optimal power values ​​of all air compressors, and the optimal total output value of the air compressor unit is obtained by summing the optimal output values ​​of all air compressors.

[0112] The adjustment power value is determined based on the comparison between the expected demand and the optimal total output;

[0113] The adjusted power value and current state label data are input into the power-output neural network model for processing to obtain the adjusted flow rate;

[0114] Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0115] The adjusted power value is input into the power-frequency neural network model for processing to obtain the adjusted frequency value. The target air compressor inverter is set to the adjusted frequency value, and the average flow rate of the adjusted air compressor unit is calculated based on the adjusted flow rate of the target air compressor.

[0116] Compare the adjusted average flow rate of the air compressor unit with the expected demand. If the comparison is consistent, the total power of the air compressor unit at this time is taken as the expected total power of the air compressor unit.

[0117] If the comparison is inconsistent, continue to adjust until the average flow rate of the adjusted air compressor unit reaches the expected demand. Then, take the total power of the air compressor unit at this time as the expected total power of the air compressor unit.

[0118] It should be noted that the efficiency corresponding to each power is calculated from the performance curves of each air compressor unit, where efficiency = output / energy consumption. The output and power of each air compressor with optimal efficiency are calculated. The output and power of all air compressors that are turned on and under load are summed to obtain the output and power of the air compressor unit with optimal efficiency. If the expected flow rate is greater than the optimal output, the air compressor with the largest increase in output when the power is increased by 0.5kW is selected for power adjustment. If the expected flow rate is less than the optimal output, the air compressor with the smallest decrease in output when the power is decreased by 0.5kW is selected for power adjustment. This process is repeated multiple times until the output reaches the expected flow rate, and the total power of the unit at this time is obtained.

[0119] Specifically, the expected demand is compared with the optimal total output. If the expected demand is greater than the optimal total output, the sum of the current power value of each air compressor and 0.5KW is used as the adjustment power value. If the expected demand is less than the optimal total output, the difference between the current power value of each air compressor and 0.5KW is used as the adjustment power value. If the expected demand is greater than the optimal total output, the air compressor with the largest absolute value of the difference is selected as the target air compressor. If the expected demand is less than the optimal total output, the air compressor with the smallest absolute value of the difference is selected as the target air compressor.

[0120] According to an embodiment of the present invention, the edge end acquires the full-load flow rate of the air compressor, compares the expected demand with the full-load flow rate, and selects the optimal scheduling optimization scheme for the air compressor unit based on the comparison result, including:

[0121] If the full-load flow rate is greater than the expected demand, calculate the expected total power of the air compressor unit with one additional air compressor and two additional air compressors respectively, and select the scheme with the lowest expected total power to be implemented.

[0122] If the full-load flow rate is less than the expected demand, calculate the expected total power and energy saving rate of the air compressor unit by adding one, removing one, and replacing one air compressor respectively, and select the scheme with the highest energy saving rate of greater than 5% to implement.

[0123] It should be noted that in this application, it is determined whether the currently operating and loaded air compressor meets the requirement that its full-load output exceeds the expected demand. If it does, the expected total power for adding one, removing one, and replacing one is calculated respectively. The scheme with the highest energy-saving rate among those that meet both the requirement of energy saving greater than 5% and full-load output exceeding the expected demand is selected for implementation. The energy-saving rate can be calculated by dividing the difference between energy consumption before and after the modification by the energy consumption before the modification. If the requirement of full-load output exceeding the expected demand is not met, the expected total power for adding one and two air compressors is calculated respectively. The scheme with the lowest total power among those that meet the requirement of full-load output exceeding the expected demand is selected for implementation.

[0124] This invention also discloses an air compressor tuning and optimization system based on artificial intelligence and cloud-edge collaboration, including a memory and a processor. The memory stores a program for an air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration. When the processor executes the program for the air compressor tuning and optimization method based on artificial intelligence and cloud-edge collaboration, it performs the following steps:

[0125] The performance model of the air compressor is modeled using a pre-defined fully connected neural network algorithm, resulting in a power-output neural network model and a power-frequency neural network model, which are then stored in the cloud.

[0126] The edge device adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model.

[0127] A demand prediction model is constructed using a pre-defined long short-term memory artificial neural network algorithm and stored in the cloud. The edge device predicts the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand.

[0128] The edge calculation determines the expected total power of the air compressor unit when it reaches the expected demand.

[0129] The full-load flow rate of the air compressor is obtained at the edge, the expected demand is compared with the full-load flow rate, and the best scheduling optimization scheme of the air compressor unit is selected based on the comparison results.

[0130] It should be noted that this application uses a fully connected neural network (FCNN) in the cloud to model the performance of the air compressor and a long short-term memory neural network (LSTM) to model the compressed air demand. At the edge, frequency conversion optimization and scheduling optimization are performed based on the performance model and demand model synchronized in the cloud, so as to ensure that the air compressor unit is always in the optimal efficiency state under the operating conditions, thereby achieving the purpose of energy saving and cost reduction.

[0131] According to an embodiment of the present invention, the step of using a preset fully connected neural network algorithm to model the performance of an air compressor, obtaining a power-output neural network model and a power-frequency neural network model, and storing them in the cloud includes:

[0132] Real-time acquisition of air compressor operating data, including instantaneous power, instantaneous frequency, status marker data, and average flow rate;

[0133] Based on a preset fully connected neural network algorithm, the instantaneous power, state marker data and average flow rate of a single air compressor are used to initialize the parameters of the preset power-output model, resulting in an initialized power-output neural network model.

[0134] The total average flow velocity of the air compressor unit is calculated based on the average flow velocity of all air compressors.

[0135] The initial power-output neural network model is trained based on the total average flow rate and the instantaneous power and state label data of each air compressor to obtain the power-output neural network model.

[0136] A power-frequency neural network model is obtained by training the model using a preset fully connected neural network algorithm and the instantaneous power and instantaneous frequency of each air compressor.

[0137] The power-output neural network model and the power-frequency neural network model are stored in the cloud.

[0138] It should be noted that the status flag data represents the start / stop and loading / unloading states as two binary features: 1 for running and loading, and 0 for stopping or unloading. The average flow rate is the average flow rate of the air compressor over 5 seconds, and the instantaneous power is input at 0.5kW intervals. First, the model parameters are initialized using data from a single running unit. Then, data from all air compressors in the same unit running simultaneously is used for model training to obtain a power-output neural network model. The total average flow rate of the air compressor unit is the sum of the average flow rates of all air compressors over 5 seconds.

[0139] According to an embodiment of the present invention, the edge terminal adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model, including:

[0140] The edge terminal acquires the current pressure value, current power value, current flow rate, and current status flag data of the air compressor;

[0141] The system power increase or decrease is obtained by performing proportional-integral calculations based on the current pressure value and the set pressure value, and the adjustment power value and adjustment number set value are determined.

[0142] The regulated power value and the current state label data are input into the power-output neural network model for processing to obtain the regulated flow rate;

[0143] Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0144] The power adjustment value is input into the power-frequency neural network model to obtain the adjustment frequency value. The target air compressor frequency converter is set to the adjustment frequency value, and the current adjustment count is updated synchronously.

[0145] The current number of adjustments is compared with the set number of adjustments. If the current number of adjustments is less than the set number of adjustments, the adjustment continues to repeat.

[0146] It should be noted that the formula for calculating proportional integrals is:

[0147] ;

[0148] ;

[0149] in, Let be the power increase or decrease at time t. This is the proportionality coefficient. The integral coefficient is... The set pressure value at time t. Let be the current pressure value at time t. For error, The integral of the error;

[0150] If the calculated value is an increase in system power, the sum of the current power value and 0.5KW is used as the adjustment power value. If the calculated value is a decrease in system power, the difference between the current power value and 0.5KW is used as the adjustment power value, and the ratio of the increase / decrease in system power to 0.5KW is used as the adjustment number. The difference between the adjusted flow rate and the current flow rate is calculated. If the calculated value is an increase in system power, the air compressor with the largest absolute value of the difference is used as the target air compressor. If the calculated value is a decrease in system power, the air compressor with the smallest absolute value of the difference is used as the target air compressor. If the proportional-integral (PI) calculation requires an increase in power, then based on the current power, the air compressor with the largest increase in output when power is increased by 0.5kW is selected, and the frequency of this air compressor is increased to achieve the increased power. If the PI calculation requires a decrease in power, then based on the current power, the air compressor with the smallest decrease in output when power is decreased by 0.5kW is selected, and the frequency of this air compressor is decreased to achieve the decreased power. This process is repeated multiple times until the power increase or decrease reaches the given power increase or decrease value, resulting in the optimized frequency for each air compressor. Each adjustment increments the current adjustment count by 1.

[0151] According to an embodiment of the present invention, the step of constructing a demand prediction model using a preset long short-term memory artificial neural network algorithm and storing it in the cloud, and then using the edge device to predict the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand, includes:

[0152] Obtain the average demand of the air compressor and the average flow rate and instantaneous pressure within a preset time period;

[0153] Based on a preset long short-term memory artificial neural network algorithm, and based on the average flow rate, instantaneous pressure and average demand within a preset time period, the model is trained to obtain a demand prediction model and store it in the cloud.

[0154] The average flow rate and instantaneous pressure of the air compressor unit within the current preset time period are obtained at the edge end;

[0155] The average flow rate and instantaneous pressure within the current preset time period are input into the preset demand prediction model to obtain the expected demand.

[0156] It should be noted that the average flow rate and instantaneous pressure within the preset time period of the air compressor refer to the average flow rate over 15 seconds and the instantaneous pressure over the previous 60 minutes before the demand forecast time. The average demand refers to the average flow rate demand over the next 5 minutes before the demand forecast time.

[0157] According to an embodiment of the present invention, the calculation of the expected total power of the air compressor unit when the expected demand is reached at the edge end includes:

[0158] The efficiency of each air compressor at different power levels is obtained at the edge, and the power value with the highest efficiency and the average flow rate are selected as the optimal power value and the optimal output value.

[0159] The optimal total power of the air compressor unit is obtained by summing the optimal power values ​​of all air compressors, and the optimal total output value of the air compressor unit is obtained by summing the optimal output values ​​of all air compressors.

[0160] The adjustment power value is determined based on the comparison between the expected demand and the optimal total output;

[0161] The adjusted power value and current state label data are input into the power-output neural network model for processing to obtain the adjusted flow rate;

[0162] Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference;

[0163] The adjusted power value is input into the power-frequency neural network model for processing to obtain the adjusted frequency value. The target air compressor inverter is set to the adjusted frequency value, and the average flow rate of the adjusted air compressor unit is calculated based on the adjusted flow rate of the target air compressor.

[0164] Compare the adjusted average flow rate of the air compressor unit with the expected demand. If the comparison is consistent, the total power of the air compressor unit at this time is taken as the expected total power of the air compressor unit.

[0165] If the comparison is inconsistent, continue to adjust until the average flow rate of the adjusted air compressor unit reaches the expected demand. Then, take the total power of the air compressor unit at this time as the expected total power of the air compressor unit.

[0166] It should be noted that the efficiency corresponding to each power is calculated from the performance curves of each air compressor unit, where efficiency = output / energy consumption. The output and power of each air compressor with optimal efficiency are calculated. The output and power of all air compressors that are turned on and under load are summed to obtain the output and power of the air compressor unit with optimal efficiency. If the expected flow rate is greater than the optimal output, the air compressor with the largest increase in output when the power is increased by 0.5kW is selected for power adjustment. If the expected flow rate is less than the optimal output, the air compressor with the smallest decrease in output when the power is decreased by 0.5kW is selected for power adjustment. This process is repeated multiple times until the output reaches the expected flow rate, and the total power of the unit at this time is obtained.

[0167] Specifically, the expected demand is compared with the optimal total output. If the expected demand is greater than the optimal total output, the sum of the current power value of each air compressor and 0.5KW is used as the adjustment power value. If the expected demand is less than the optimal total output, the difference between the current power value of each air compressor and 0.5KW is used as the adjustment power value. If the expected demand is greater than the optimal total output, the air compressor with the largest absolute value of the difference is selected as the target air compressor. If the expected demand is less than the optimal total output, the air compressor with the smallest absolute value of the difference is selected as the target air compressor.

[0168] According to an embodiment of the present invention, the edge end acquires the full-load flow rate of the air compressor, compares the expected demand with the full-load flow rate, and selects the optimal scheduling optimization scheme for the air compressor unit based on the comparison result, including:

[0169] If the full-load flow rate is greater than the expected demand, calculate the expected total power of the air compressor unit with one additional air compressor and two additional air compressors respectively, and select the scheme with the lowest expected total power to be implemented.

[0170] If the full-load flow rate is less than the expected demand, calculate the expected total power and energy saving rate of the air compressor unit by adding one, removing one, and replacing one air compressor respectively, and select the scheme with the highest energy saving rate of greater than 5% to implement.

[0171] It should be noted that in this application, it is determined whether the currently operating and loaded air compressor meets the requirement that its full-load output exceeds the expected demand. If it does, the expected total power for adding one, removing one, and replacing one is calculated respectively. The scheme with the highest energy-saving rate among those that meet both the requirement of energy saving greater than 5% and full-load output exceeding the expected demand is selected for implementation. The energy-saving rate can be calculated by dividing the difference between energy consumption before and after the modification by the energy consumption before the modification. If the requirement of full-load output exceeding the expected demand is not met, the expected total power for adding one and two air compressors is calculated respectively. The scheme with the lowest total power among those that meet the requirement of full-load output exceeding the expected demand is selected for implementation.

[0172] The invention discloses an air compressor optimization method and system based on artificial intelligence and cloud-edge collaboration. By using a fully connected neural network (FCNN) to model the air compressor performance in the cloud and a long short-term memory neural network (LSTM) to model the compressed air demand, the system performs frequency conversion optimization and scheduling optimization at the edge based on the performance model and demand model synchronized in the cloud. This ensures that the air compressor unit is always in the optimal efficiency state under the operating conditions, thereby achieving the goal of energy saving and cost reduction.

[0173] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0174] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0175] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

Claims

1. An air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration, characterized in that, Includes the following steps: The performance model of the air compressor is modeled using a pre-defined fully connected neural network algorithm, resulting in a power-output neural network model and a power-frequency neural network model, which are then stored in the cloud. The edge device adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model. A demand prediction model is constructed using a pre-defined long short-term memory artificial neural network algorithm and stored in the cloud. The edge device predicts the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand. The edge calculation determines the expected total power of the air compressor unit when it reaches the expected demand. The full-load flow rate of the air compressor is obtained at the edge, the expected demand is compared with the full-load flow rate, and the best scheduling optimization scheme of the air compressor unit is selected based on the comparison results.

2. The air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration according to claim 1, characterized in that, The process involves using a pre-defined fully connected neural network algorithm to model the air compressor performance, obtaining a power-output neural network model and a power-frequency neural network model, and storing them in the cloud, including: Real-time acquisition of air compressor operating data, including instantaneous power, instantaneous frequency, status marker data, and average flow rate; Based on a preset fully connected neural network algorithm, the instantaneous power, state marker data and average flow rate of a single air compressor are used to initialize the parameters of the preset power-output model, resulting in an initialized power-output neural network model. The total average flow velocity of the air compressor unit is calculated based on the average flow velocity of all air compressors. The initial power-output neural network model is trained based on the total average flow rate and the instantaneous power and state label data of each air compressor to obtain the power-output neural network model. A power-frequency neural network model is obtained by training the model using a preset fully connected neural network algorithm and the instantaneous power and instantaneous frequency of each air compressor. The power-output neural network model and the power-frequency neural network model are stored in the cloud.

3. The air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration according to claim 2, characterized in that, The edge terminal adjusts the frequency of the air compressor according to the power-output neural network model and the power-frequency neural network model, including: The edge terminal acquires the current pressure value, current power value, current flow rate, and current status flag data of the air compressor; The system power increase or decrease is obtained by performing proportional-integral calculations based on the current pressure value and the set pressure value, and the adjustment power value and adjustment number set value are determined. The regulated power value and the current state label data are input into the power-output neural network model for processing to obtain the regulated flow rate; Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference; The power adjustment value is input into the power-frequency neural network model to obtain the adjustment frequency value. The target air compressor frequency converter is set to the adjustment frequency value, and the current adjustment count is updated synchronously. The current number of adjustments is compared with the set number of adjustments. If the current number of adjustments is less than the set number of adjustments, the adjustment continues to repeat.

4. The air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration according to claim 3, characterized in that, The process involves constructing a demand prediction model using a pre-defined long short-term memory artificial neural network algorithm and storing it in the cloud. The edge device then predicts the current demand of the air compressor unit based on this model to obtain the expected demand, including: Obtain the average demand of the air compressor and the average flow rate and instantaneous pressure within a preset time period; Based on a preset long short-term memory artificial neural network algorithm, and based on the average flow rate, instantaneous pressure and average demand within a preset time period, the model is trained to obtain a demand prediction model and store it in the cloud. The average flow rate and instantaneous pressure of the air compressor unit within the current preset time period are obtained at the edge end; The average flow rate and instantaneous pressure within the current preset time period are input into the preset demand prediction model to obtain the expected demand.

5. The air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration according to claim 4, characterized in that, The edge-end calculation of the expected total power of the air compressor unit when it reaches the expected demand includes: The efficiency of each air compressor at different power levels is obtained at the edge, and the power value with the highest efficiency and the average flow rate are selected as the optimal power value and the optimal output value. The optimal total power of the air compressor unit is obtained by summing the optimal power values ​​of all air compressors, and the optimal total output value of the air compressor unit is obtained by summing the optimal output values ​​of all air compressors. The adjustment power value is determined based on the comparison between the expected demand and the optimal total output; The adjusted power value and current state label data are input into the power-output neural network model for processing to obtain the adjusted flow rate; Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference; The adjusted power value is input into the power-frequency neural network model for processing to obtain the adjusted frequency value. The target air compressor inverter is set to the adjusted frequency value, and the average flow rate of the adjusted air compressor unit is calculated based on the adjusted flow rate of the target air compressor. Compare the adjusted average flow rate of the air compressor unit with the expected demand. If the comparison is consistent, the total power of the air compressor unit at this time is taken as the expected total power of the air compressor unit. If the comparison is inconsistent, continue to adjust until the average flow rate of the adjusted air compressor unit reaches the expected demand. Then, take the total power of the air compressor unit at this time as the expected total power of the air compressor unit.

6. The air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration according to claim 5, characterized in that, The edge terminal acquires the full-load flow rate of the air compressor, compares the expected demand with the full-load flow rate, and selects the optimal scheduling optimization scheme for the air compressor unit based on the comparison result, including: If the full-load flow rate is less than the expected demand, calculate the expected total power of the air compressor unit with one additional air compressor and two additional air compressors respectively, and select the scheme with the lowest expected total power to be implemented. If the full-load flow rate is greater than the expected demand, calculate the expected total power and energy saving rate of the air compressor unit by reducing one and replacing one air compressor respectively, and select the scheme with the highest energy saving rate of greater than 5% to implement.

7. An air compressor scheduling and optimization system based on artificial intelligence and cloud-edge collaboration, characterized in that, The system includes a memory and a processor. The memory stores a program for an air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration. When the program for the air compressor scheduling optimization method based on artificial intelligence and cloud-edge collaboration is executed by the processor, it performs the following steps: The performance model of the air compressor is modeled using a pre-defined fully connected neural network algorithm, resulting in a power-output neural network model and a power-frequency neural network model, which are then stored in the cloud. The edge device adjusts the frequency of the air compressor based on the power-output neural network model and the power-frequency neural network model. A demand prediction model is constructed using a pre-defined long short-term memory artificial neural network algorithm and stored in the cloud. The edge device predicts the current demand of the air compressor unit based on the demand prediction model to obtain the expected demand. The edge calculation determines the expected total power of the air compressor unit when it reaches the expected demand. The full-load flow rate of the air compressor is obtained at the edge, the expected demand is compared with the full-load flow rate, and the best scheduling optimization scheme of the air compressor unit is selected based on the comparison results.

8. The air compressor scheduling optimization system based on artificial intelligence and cloud-edge collaboration according to claim 7, characterized in that, The process involves using a pre-defined fully connected neural network algorithm to model the air compressor performance, obtaining a power-output neural network model and a power-frequency neural network model, and storing them in the cloud, including: Real-time acquisition of air compressor operating data, including instantaneous power, instantaneous frequency, status marker data, and average flow rate; Based on a preset fully connected neural network algorithm, the instantaneous power, state marker data and average flow rate of a single air compressor are used to initialize the parameters of the preset power-output model, resulting in an initialized power-output neural network model. The total average flow velocity of the air compressor unit is calculated based on the average flow velocity of all air compressors. The initial power-output neural network model is trained based on the total average flow rate and the instantaneous power and state label data of each air compressor to obtain the power-output neural network model. A power-frequency neural network model is obtained by training the model using a preset fully connected neural network algorithm and the instantaneous power and instantaneous frequency of each air compressor. The power-output neural network model and the power-frequency neural network model are stored in the cloud.

9. The air compressor scheduling optimization system based on artificial intelligence and cloud-edge collaboration according to claim 8, characterized in that, The edge terminal adjusts the frequency of the air compressor according to the power-output neural network model and the power-frequency neural network model, including: The edge terminal acquires the current pressure value, current power value, current flow rate, and current status flag data of the air compressor; The system power increase or decrease is obtained by performing proportional-integral calculations based on the current pressure value and the set pressure value, and the adjustment power value and adjustment number set value are determined. The regulated power value and the current state label data are input into the power-output neural network model for processing to obtain the regulated flow rate; Calculate the difference between the adjusted flow rate and the current flow rate, and select the target air compressor based on the difference; The power adjustment value is input into the power-frequency neural network model to obtain the adjustment frequency value. The target air compressor frequency converter is set to the adjustment frequency value, and the current adjustment count is updated synchronously. The current number of adjustments is compared with the set number of adjustments. If the current number of adjustments is less than the set number of adjustments, the adjustment continues to repeat.

10. The air compressor scheduling optimization system based on artificial intelligence and cloud-edge collaboration according to claim 9, characterized in that, The process involves constructing a demand prediction model using a pre-defined long short-term memory artificial neural network algorithm and storing it in the cloud. The edge device then predicts the current demand of the air compressor unit based on this model to obtain the expected demand, including: Obtain the average demand of the air compressor and the average flow rate and instantaneous pressure within a preset time period; Based on a preset long short-term memory artificial neural network algorithm, and based on the average flow rate, instantaneous pressure and average demand within a preset time period, the model is trained to obtain a demand prediction model and store it in the cloud. The average flow rate and instantaneous pressure of the air compressor unit within the current preset time period are obtained at the edge end; The average flow rate and instantaneous pressure within the current preset time period are input into the preset demand prediction model to obtain the expected demand.

Images