Air pressure regulation and control system for multiple oil-free air compression stations

Abstract

The invention relates to an air pressure regulation and control system for multiple oil-free air compression stations. The air pressure regulation and control system is characterized in that the current air pressure, air production flow, air consumption equipment operation state and air consumption flow of the multiple air compression stations are monitored; establishing a deep learning model of the operation state and the gas consumption flow of the gas consumption equipment, and predicting the gas consumption flow of the next time period according to the current operation state and the current gas consumption flow of the gas consumption equipment; and according to the air consumption flow of the next time period, the number of the operating air compressors and the air consumption flow are adjusted, so that the preset air pressure range is kept. The problems that in the prior art, complex and changeable industrial gas use scenes are difficult to deal with, the operation parameters of the air compressor cannot be adjusted timely and accurately, and the air pressure is prone to being unstable are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of air compressor stations, and relates to a pneumatic pressure regulation system for multiple oil-free air compressor stations. Background Art

[0002] In the process of industrial production, compressed air, as a key power source, its pressure stability plays a decisive role in production efficiency and product quality. Oil-free air compressor stations, with the characteristic of outputting clean compressed air, are widely used in industrial fields with strict requirements for air quality. With the expansion of industrial scale and the refined development of production processes, a single oil-free air compressor can no longer meet the growing gas consumption demand, and the application of multiple oil-free air compressor stations is becoming more and more common. However, when multiple air compressors operate in coordination, there are many challenges in pneumatic pressure regulation, such as pressure fluctuations and energy waste. An effective pneumatic pressure regulation system is the key to the stable and efficient operation of multiple oil-free air compressor stations. It can not only ensure that the air pressure is accurately maintained within the set range to meet the gas consumption requirements of different production equipment, but also achieve energy conservation and consumption reduction through optimized control strategies and extend the service life of the equipment. Therefore, in-depth research on the pneumatic pressure regulation strategy for multiple oil-free air compressor stations is of great significance for improving the comprehensive benefits of industrial production and promoting the sustainable development of the industry.

[0003] However, the traditional pneumatic pressure regulation of multiple oil-free air compressor stations often relies on preset rules and fixed parameters to operate. These rules are usually established based on experience or some simple mathematical models and are difficult to cope with complex and changeable industrial gas consumption scenarios. For example, different production processes, different time periods, and different operating states of equipment will all lead to large fluctuations in gas consumption. In this case, the traditional system cannot adjust the operating parameters of the air compressor in a timely and accurate manner, easily causing unstable air pressure, affecting production quality, and may also lead to energy waste. Summary of the Invention

[0004] To solve the above problems existing in the prior art, the present invention provides a pneumatic pressure regulation system for multiple oil-free air compressor stations, aiming to deeply analyze a large amount of data such as air pressure, flow rate, and equipment operating status collected in real time by sensors through a deep learning model. Through continuous learning and training, accurately predict the gas consumption in different time periods and working conditions, and adjust the operating parameters and the number of operating air compressors in advance to achieve forward-looking and intelligent regulation of air pressure.

[0005] The object of the present invention can be achieved by the following technical solutions:

[0006] The present application provides a pneumatic pressure regulation system for multiple oil-free air compressor stations, including a working monitoring module, a gas consumption estimation module, and a pneumatic pressure regulation module. The working monitoring module, the gas consumption estimation module, and the pneumatic pressure regulation module are communicatively connected, where:

[0007] The working monitoring module is used to monitor the current air pressure, gas production flow rate, operating status of gas-using equipment, and gas consumption flow rate of multiple air compressor stations;

[0008] The gas consumption prediction module is used to establish a deep learning model for the operating status and gas consumption flow rate of gas-using equipment, and predict the gas consumption flow rate in the next time period based on the current operating status and gas consumption flow rate of gas-using equipment;

[0009] The air pressure regulation module is used to adjust the number of operating air compressors and the gas consumption flow rate according to the gas consumption flow rate in the next time period to maintain a preset air pressure range.

[0010] Further, the operating status of the gas-using equipment includes start-stop status, load status, working pressure, working flow rate, and fault status.

[0011] Further, the deep learning model includes a time series model and a random forest model. The time series model is used to predict the operating status of gas-using equipment and the first gas consumption flow rate in the next time period based on the current operating status and gas consumption flow rate of gas-using equipment; the random forest model is used to predict the second gas consumption flow rate in the next time period based on the operating status of gas-using equipment in the next time period.

[0012] Further, the gas consumption flow rate in the next time period is obtained by comprehensively calculating the first gas consumption flow rate and the second gas consumption flow rate.

[0013] Further, the comprehensive calculation by the first gas consumption flow rate and the second gas consumption flow rate is as follows: the calculation formula is:

[0014] Q Z = w1Q1 + w2Q2,

[0015] In the formula, Q Z is the gas consumption flow rate in the next time period; Q1 is the first gas consumption flow rate, w1 is the corresponding first weight; Q2 is the second gas consumption flow rate, w2 is the corresponding second weight; the first weight and the second weight are determined by the average error predicted by the model in the historical period.

[0016] Further, the time series model is configured as a long short-term memory network model, including the following construction steps:

[0017] Data collection and integration: Collect historical data covering the operating status and gas consumption flow rate of gas-using equipment from multiple air compressor stations in the working monitoring module, and process missing values and outliers;

[0018] Data preprocessing: Normalize the data, organize it in a specific format according to time order, and determine the time step to construct a time series data set;

[0019] Model architecture design: Determine the number of neurons in the input layer according to the data characteristics, select an appropriate number of LSTM units to construct the LSTM layer, and set the number of neurons in the output layer according to the prediction target;

[0020] Model compilation: Select the mean squared error as the loss function to measure the error, use Adam as the optimizer to update the model parameters and set the learning rate;

[0021] Model training: Divide the training set, validation set, and test set. The training set is used to train the model, set the number of training epochs and batch size for training, and the validation set is used for model validation and optimization of the goodness of fit;

[0022] Model evaluation and tuning: Use the test set to evaluate and judge the prediction performance of the model, and optimize the model by adjusting the hyperparameters and increasing the data samples.

[0023] Further, the random forest model includes the following construction steps:

[0024] Data preparation: Collect data on the operating status of gas-using equipment and the corresponding gas flow rates from the work monitoring module, and select the key features that have a significant impact on gas flow rate prediction through the correlation analysis method;

[0025] Dataset division: Divide the preprocessed data into a training set, a validation set, and a test set;

[0026] Model initialization: Determine the number of decision trees in the random forest model, and set the number of randomly selected features, maximum depth, and minimum sample split number for each tree;

[0027] Model training: Use the training set data, through the bootstrap sampling method with replacement, to construct unique training samples for each decision tree; When each tree divides nodes, randomly select some features and train independently. Finally, integrate the prediction results of all decision trees to complete the training process of the random forest model;

[0028] Model evaluation: First, use the validation set to preliminarily evaluate the trained model and adjust the hyperparameters to optimize the model; Then use the test set for the final evaluation to measure the generalization ability and prediction accuracy of the model on unknown data.

[0029] Further, adjusting the number of operating air compressors and the gas flow rate according to the gas flow rate in the next time period to maintain the preset air pressure range includes the following steps:

[0030] Gas flow analysis: Obtain the predicted gas flow rate data for the next time period from the gas consumption estimation module, combine it with the gas production capacity of the current air compressor and the real-time air pressure state of the system to evaluate the supply and demand state of the system;

[0031] Air compressor operation unit adjustment decision: Preset the threshold of the number of air compressors required to operate based on different gas production scales. Compare the result obtained from the analysis of gas consumption flow with the preset unit threshold. If the predicted gas consumption flow increases significantly and the current gas production capacity of the air compressor is difficult to match, increase the number of operating air compressors; conversely, if the predicted gas consumption flow decreases and the current gas production capacity is significantly excessive, reduce the number of operating units;

[0032] Specific air compressor start / stop decision: Develop a priority list for air compressors, and determine the priority of each air compressor by comprehensively considering equipment performance, service life, and maintenance records; According to the operation unit adjustment decision, combined with the priority list, determine the specific air compressors that need to be started or stopped;

[0033] Air compressor operation parameter adjustment: For the operating or about-to-start air compressors, adjust the operation parameters according to the predicted gas consumption flow. Based on the pre-set corresponding relationship between the gas production flow of the air compressor and the motor speed, calculate the required motor speed according to the predicted gas consumption flow, and then send an instruction to the frequency converter for adjustment;

[0034] Real-time monitoring and feedback adjustment: After completing the adjustment of the number of operating air compressors and parameters, conduct real-time and high-frequency monitoring of the actual air pressure and gas consumption flow. Compare the real-time obtained data with the preset air pressure range. If the actual air pressure exceeds or approaches the boundary of the preset range, fine-tune the operation parameters or the number of operating units of the air compressor again according to the deviation degree;

[0035] Recording and reporting: Record each adjustment operation of the number of operating air compressors and gas consumption flow, covering the adjustment time, adjustment reason, air compressor numbers involved, and specific adjustment parameters.

[0036] Furthermore, the air pressure control module uses a distributed cooperative control architecture to coordinate the control of multiple air compressor stations.

[0037] Advantages of the present invention:

[0038] By monitoring the current air pressure, gas production flow, operating status of gas-consuming equipment, and gas consumption flow of multiple air compressor stations; establishing a deep learning model for the operating status and gas consumption flow of gas-consuming equipment, predicting the gas consumption flow in the next time period according to the current operating status and gas consumption flow of gas-consuming equipment; adjusting the number of operating air compressors and gas consumption flow according to the gas consumption flow in the next time period to maintain the preset air pressure range. The present invention solves the problems that the prior art is difficult to cope with complex and changeable industrial gas consumption scenarios, cannot adjust the operation parameters of air compressors in a timely and accurate manner, and is prone to unstable air pressure. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings.

[0040] Figure 1 This is a structural diagram of a pneumatic pressure regulation system for multiple oil-free air compressor stations in the present invention.

[0041] Figure 2 This is a flowchart for adjusting the number of operating air compressors and the gas consumption flow rate according to the gas consumption flow rate in the next time period in an embodiment of the present invention to maintain a preset air pressure range. Detailed implementation manners

[0042] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, details the specific implementation manners, structures, features, and their effects according to the present invention as follows.

[0043] Please refer to Figure 1 - Figure 2 , this application provides a pneumatic pressure regulation system for multiple oil-free air compressor stations, including a working monitoring module, a gas consumption prediction module, and a pneumatic pressure regulation module. The working monitoring module, the gas consumption prediction module, and the pneumatic pressure regulation module are communicatively connected, wherein:

[0044] The working monitoring module is used to monitor the current air pressure, gas production flow rate, operating status of gas-using equipment, and gas consumption flow rate of multiple air compressor stations;

[0045] In this embodiment, in the pneumatic pressure regulation system for multiple oil-free air compressor stations, the working monitoring module provides basic data for the stable operation of the system. This module monitors the current air pressure of multiple air compressor stations. Air pressure is a key indicator for the operation of air compressor stations and is related to whether the gas-using equipment can work properly. At key parts of the air compressor station, such as the main pipe, the air outlet of the air compressor, and the front end of the gas-using equipment, pressure sensors are installed, and the working monitoring module uses them to collect air pressure data in real time. These data can be used to detect abnormal air pressure, avoid equipment damage caused by improper air pressure, and at the same time provide a basis for air pressure regulation. The working monitoring module is also responsible for monitoring the gas production flow rate. The gas production flow rate of the air compressor reflects its working efficiency and operating status. With the help of flow sensors, this module obtains the gas production flow rate information of each air compressor in real time, thereby judging whether the air compressor is working properly and whether there are abnormal gas production situations. By integrating the gas production flow rate data of each unit, it can provide a reference for reasonably arranging the number of operating air compressors and achieve efficient energy utilization. In addition, the working monitoring module monitors the operating status and gas consumption flow rate of gas-using equipment. Different gas-using equipment has different gas consumption demands under different operating states. By monitoring the operating states of gas-using equipment such as startup, stop, and load status, combined with real-time gas consumption flow rate data, the gas consumption demand of the air compressor station can be grasped. This is of great significance for the gas consumption prediction module to build a deep learning model and the pneumatic pressure regulation module to achieve precise pneumatic pressure regulation. The working monitoring module comprehensively and real-time monitors the air pressure, gas production flow rate, operating status of gas-using equipment, and gas consumption flow rate of multiple air compressor stations, provides data support for the pneumatic pressure regulation system for multiple oil-free air compressor stations, and is the key to the efficient and stable operation of the system.

[0046] Further, the operating state of the air-using device includes start-stop state, load state, working pressure, working flow rate, and fault state.

[0047] In this embodiment, the start-stop state directly reflects whether the air-using device is connected to the air compressor system and starts consuming compressed air. When the device starts, it will instantaneously generate an air demand, which may cause a certain degree of decrease in air pressure; when the device stops running, the air demand disappears, and the air pressure may increase. Accurately monitoring the start-stop state helps the system to anticipate the sudden change in air flow rate in a timely manner, adjust the operating state of the air compressor in advance, and ensure the stability of air pressure. For example, on an automated production line, multiple pneumatic devices start and stop in sequence, and the system reasonably arranges the loading and unloading of the air compressor according to the start-stop signals to avoid large fluctuations in air pressure.

[0048] The load state reflects the working intensity of the air-using device. The air demand of the device varies greatly under different loads. At high loads, the device requires more compressed air to complete the corresponding work tasks, and the requirement for air pressure stability is also higher; at low loads, the air demand is relatively small. Understanding the load state of the device can enable the system to adjust the air supply strategy more precisely, meet the air demand of the device while avoiding energy waste. For example, when a pneumatic processing device is performing rough machining and finish machining, the loads are different, and the system can adjust the gas production flow rate of the air compressor according to the load change.

[0049] The working pressure is one of the key parameters for the normal operation of the air-using device. Each air-using device has its specific working pressure range, and only within this range can the device work efficiently and stably. If the air pressure is lower than the working pressure required by the device, the device may not be able to start normally or the operating efficiency may decrease; if the air pressure is too high, it may damage the device components. Therefore, real-time monitoring of the working pressure and feeding it back to the air pressure control system helps to adjust the output pressure of the air compressor in a timely manner to ensure that the device works in a suitable pressure environment. For example, some precision pneumatic measuring instruments have extremely high requirements for the accuracy of working pressure, and the system needs to precisely control the air pressure to meet their working requirements.

[0050] The working flow rate represents the amount of compressed air consumed by the air-using device per unit time. It is closely related to the working state and load of the device. By monitoring the working flow rate, the system can understand the actual air demand of the device and reasonably allocate the gas production flow rate of the air compressor. The working flow rates of different air-using devices vary greatly. For example, the working flow rate of a large pneumatic stirring device is usually much larger than that of a small pneumatic spray gun. The system dynamically adjusts the number of operating air compressors and the gas production capacity according to the change in the working flow rate to achieve precise control of air pressure.

[0051] The fault status reflects whether there are abnormal conditions in the gas-using equipment. When a fault occurs in the equipment, its gas consumption demand and operating status may change, and may even affect the stability of the entire air compressor system. For example, a blocked gas path may cause a decrease in the gas flow rate of the equipment and at the same time cause a local increase in air pressure; a leak in the equipment will cause the air pressure to drop and increase the additional gas consumption demand. Timely detecting the fault status of the equipment and feeding back the information to the air pressure control system and maintenance personnel helps to quickly eliminate the fault, restore the normal operation of the system, and ensure the stable supply of air pressure.

[0052] The gas consumption prediction module is used to establish a deep learning model for the operating status and gas flow rate of the gas-using equipment, and predict the gas flow rate in the next time period according to the current operating status and gas flow rate of the gas-using equipment;

[0053] Furthermore, the deep learning model includes a time series model and a random forest model. The time series model is used to predict the operating status of the gas-using equipment and the first gas flow rate in the next time period according to the current operating status and gas flow rate of the gas-using equipment; the random forest model is used to predict the second gas flow rate in the next time period according to the operating status of the gas-using equipment in the next time period.

[0054] In this embodiment, the deep learning model is composed of a time series model and a random forest model, and they play different but cooperative roles in predicting the gas flow rate in the next time period. The time series model is mainly based on the data analysis in the time dimension. It uses the currently obtained operating status and gas flow rate data of the gas-using equipment for prediction. Since the operation of the gas-using equipment often has certain periodicity and trend, for example, certain production processes will regularly start or stop specific equipment within a fixed time period, resulting in corresponding changes in the gas flow rate. The time series model can capture these characteristics and laws that change with time, and through the learning and analysis of historical data, infer the possible operating status of the gas-using equipment in the next time period and the corresponding first gas flow rate.

[0055] The random forest model focuses on using the operating status of gas-using equipment in the next time period to predict the second gas flow rate. The random forest is an ensemble learning method composed of multiple decision trees. In this scenario, the random forest model comprehensively considers various factors of the operating status of gas-using equipment and their interrelationships. Different combinations of the operating status of gas-using equipment correspond to different gas flow rates. Through training on a large amount of historical data, the random forest model learns the complex mapping relationship between these statuses and the gas flow rate. When the time series model predicts the operating status of gas-using equipment in the next time period, the random forest model can further accurately predict the second gas flow rate in that time period based on this status information. By combining these two models, the gas flow rate in the next time period can be predicted more comprehensively and accurately, providing a more reliable basis for the air pressure regulation module, so as to more precisely adjust the number of operating air compressors and the gas flow rate, and maintain the air pressure in the air compressor station within the preset range.

[0056] Furthermore, the gas flow rate in the next time period is calculated by comprehensively calculating the first gas flow rate and the second gas flow rate.

[0057] In this embodiment, in the air pressure regulation system of multiple oil-free air compressor stations, the first gas flow rate predicted by the time series model and the second gas flow rate predicted by the random forest model each have their own focuses. The time series model predicts based on the operating status of gas-using equipment and the time law of gas flow rate, and can reflect the change trend of gas flow rate over time and certain periodic characteristics. The random forest model, on the other hand, focuses on the complex mapping relationship between the operating status of gas-using equipment and the gas flow rate, and has good predictive ability for the gas flow rate under different combinations of operating statuses. By synthesizing these two prediction results, their advantages can be fully utilized, reducing the deviation that may be brought by single-model prediction, and making the prediction of the gas flow rate in the next time period more accurate and comprehensive.

[0058] Furthermore, the calculation by comprehensively calculating the first gas flow rate and the second gas flow rate has the following calculation formula:

[0059] Q Z = w1Q1 + w2Q2,

[0060] In the formula, Q Z is the gas flow rate in the next time period; Q1 is the first gas flow rate, w1 is the corresponding first weight; Q2 is the second gas flow rate, w2 is the corresponding second weight; the first weight and the second weight are determined by the average error predicted by the model in the historical period.

[0061] Furthermore, the time series model is configured as a long short-term memory network model, including the following construction steps:

[0062] Data collection and integration: Comprehensively obtain historical data from the work monitoring module. This data includes information such as air pressure, gas production flow rate, start-stop status, load status, working pressure, working flow rate, and fault status of multiple air compressor stations at different time periods. Ensure that the data has sufficient time span and diversity to reflect various production conditions and time patterns. Clean the collected data, check and handle missing values and outliers. For a small number of missing values, methods such as linear interpolation or mean filling can be used; for outliers, identify and correct them using statistical methods (such as based on standard deviation) according to the characteristics of the data distribution.

[0063] Data preprocessing: Normalize various types of data so that their value ranges are between [0,1] or [-1,1], eliminate the influence of different feature dimensions, and accelerate the model convergence speed. According to the time sequence, organize the data into a format suitable for input to the Long Short-Term Memory (LSTM) model. Determine an appropriate time step, and use the data of the past several time steps as input and the data of the next time step as output.

[0064] Model architecture design:

[0065] Input layer: Determine the number of neurons in the input layer according to the number of data features after preprocessing. LSTM layer: Select an appropriate number of LSTM units to construct the LSTM layer. LSTM units can effectively capture long-term dependencies in time series. Different numbers of units can be tried according to the complexity of the problem and the amount of data, and the optimal setting can be determined through experiments. At the same time, multiple LSTM layers can be considered stacked to increase the model's expressive ability, but overfitting should be avoided.

[0066] Output layer: Determine the number of neurons in the output layer according to the prediction target.

[0067] Model compilation:

[0068] Loss function: Select an appropriate loss function to measure the error between the model's predicted value and the true value. For gas flow prediction, the mean squared error (MSE) is commonly used as the loss function; for the prediction of the operating status of gas-using equipment, the cross-entropy loss function can be used (if it is a classification problem).

[0069] Optimizer: Select an optimization algorithm to update the model's parameters. Commonly used optimizers include Adam, RMSProp, etc. Set an appropriate learning rate. If the learning rate is too large, the model may not converge; if it is too small, the training speed will be slow.

[0070] Model training:

[0071] Dataset division: Divide the preprocessed time series dataset into a training set, a validation set, and a test set. Generally, the ratio can be set to 70%, 15%, 15%.

[0072] Training the model: Use the training set to train the LSTM model, and set appropriate number of training epochs and batch size. In each training epoch, the model calculates the predicted values through forward propagation, calculates the error according to the loss function, and then updates the model's parameters through the backpropagation algorithm. At the same time, after each training epoch, use the validation set to evaluate the performance of the model, observe the change of the loss function, and prevent overfitting.

[0073] Model evaluation and tuning:

[0074] Evaluation metrics: Use the test set to evaluate the trained model, and calculate relevant evaluation metrics such as mean squared error (MSE), mean absolute error (MAE), mean absolute percentage error (MAPE), etc., to comprehensively evaluate the prediction accuracy and generalization ability of the model.

[0075] Model tuning: If the model performance is not ideal, try to adjust the hyperparameters of the model, such as the number of LSTM units, learning rate, time step, etc.; you can also try to increase the training data, improve the data preprocessing method, or adopt regularization techniques (such as Dropout) to prevent overfitting and improve the performance of the model.

[0076] Furthermore, the random forest model includes the following construction steps:

[0077] Data preparation:

[0078] Data collection: Obtain relevant data from the work monitoring module, covering various operating states of gas-using equipment (start-stop state, load state, working pressure, working flow, fault state, etc.) and the corresponding gas flow data. At the same time, collect data under different time periods and different production conditions to ensure the comprehensiveness and diversity of the data.

[0079] Data cleaning: Check whether there are missing values and outliers in the data. For missing values, you can use mean, median filling or prediction filling according to other relevant features; for outliers, identify and process them by setting reasonable thresholds or using statistical methods (such as the Z-score method), and you can choose to correct or directly remove them.

[0080] Feature selection: Analyze the correlation between each feature and the gas flow, and select the features that have a greater impact on the gas flow prediction. Avoid introducing too many irrelevant or redundant features to reduce the complexity and computational amount of the model, and at the same time improve the accuracy and generalization ability of the model.

[0081] Dataset division:

[0082] Divide the processed data into a training set, a validation set, and a test set according to a certain ratio. The training set is used for the model to learn the patterns and regularities in the data; the validation set is used to adjust the hyperparameters during the model training process, evaluate the performance of the model under different hyperparameter combinations, and prevent overfitting; the test set is used to finally evaluate the generalization ability and prediction accuracy of the model.

[0083] Model initialization:

[0084] Determine the number of decision trees: A random forest is an ensemble model composed of multiple decision trees, and the number of decision trees (n_estimators) needs to be determined. Generally, through experiments, different numbers of decision trees (such as 50, 100, 200, etc.) can be tried, observe the performance of the model on the validation set, and select the number of decision trees corresponding to the best performance.

[0085] Set other parameters: Some other hyperparameters also need to be set, such as the number of features randomly selected when building each decision tree (max_features), the maximum depth of the decision tree (max_depth), the minimum number of samples for splitting (min_samples_split), etc. These parameters will affect the structure of the decision tree and the performance of the model, and methods such as grid search and random search can be used for optimization.

[0086] Model training:

[0087] Use the data in the training set to train the random forest model. During the training process, the random forest will randomly draw samples from the training set with replacement (bootstrap sampling) to build each decision tree. At the same time, a part of the features will be randomly selected at each node for splitting to increase the randomness and diversity of the model. Each decision tree is trained independently to learn the relationship between the features and labels in the data.

[0088] After training, each decision tree can make predictions on the input data, and the final prediction result of the random forest is the combination of the prediction results of all decision trees.

[0089] Model evaluation:

[0090] Use the validation set to conduct a preliminary evaluation of the trained model, calculate relevant evaluation metrics, such as mean squared error (MSE), mean absolute error (MAE), coefficient of determination (R 2 ) etc. According to the evaluation results, adjust the hyperparameters of the model to further optimize the performance of the model.

[0091] Finally, use the test set to conduct a final evaluation of the tuned model to obtain the true performance of the model on unknown data and ensure that the model has good generalization ability.

[0092] The air pressure control module is used to adjust the number of operating air compressors and the air flow rate according to the air consumption flow rate in the next time period, so as to maintain a preset air pressure range.

[0093] Further, adjusting the number of operating air compressors and the air flow rate according to the air consumption flow rate in the next time period to maintain a preset air pressure range includes the following steps:

[0094] Air flow rate analysis: Receive the air consumption flow rate data predicted by the air consumption prediction module for the next time period. Analyze this air flow rate data to judge its relationship with the current air production capacity of the air compressor and the system air pressure state. For example, calculate the difference between the predicted air consumption flow rate and the total current air production flow rate of the air compressor to clarify whether the air production is sufficient.

[0095] Decision-making on adjusting the number of operating air compressors: Set the threshold values of the number of operating air compressors required under different air production scales. For example, when the predicted air consumption flow rate is in a certain interval, a specific number of air compressors are correspondingly started. Compare the result of the air flow rate analysis with the preset threshold values of the number of units. If the predicted air consumption flow rate increases and the current air production capacity of the air compressor is insufficient to meet the demand, decide to increase the number of operating air compressors; if the predicted air consumption flow rate decreases and the current air production capacity is excessive, consider reducing the number of operating air compressors.

[0096] Decision-making on starting and stopping air compressors: Establish a priority list of air compressors, and determine the priority according to factors such as the equipment performance, service life, and maintenance status of the air compressors. For example, give priority to starting air compressors with better performance and good maintenance records; when stopping is required, give priority to stopping air compressors with a longer service life or higher recent maintenance requirements. Determine the specific air compressors that need to be started or stopped according to the decision-making on adjusting the number of operating air compressors and in combination with the priority list.

[0097] Adjustment of air compressor operating parameters: For the air compressors that are running or about to start, further adjust their operating parameters according to the predicted air consumption flow rate. For example, adjust the motor speed of the air compressor through a frequency converter, thereby changing the air production flow rate. Set the corresponding relationship between the air production flow rate of the air compressor and the motor speed, calculate the required motor speed according to the predicted air consumption flow rate, and send an instruction to the frequency converter for adjustment.

[0098] Real-time monitoring and feedback adjustment: After adjusting the number of operating air compressors and parameters, monitor the actual air pressure and air consumption flow rate of the system in real time. Obtain real-time data through the pressure sensors and flow sensors installed in the system. Compare the real-time monitored air pressure and air consumption flow rate data with the preset air pressure range. If the actual air pressure exceeds or approaches the boundary of the preset range, fine-tune the operating parameters or the number of operating air compressors of the air compressor again according to the deviation degree to ensure that the air pressure is always maintained within the preset range.

[0099] Recording and Reporting: Record each adjustment operation of the number of operating air compressors and gas consumption flow rate, including information such as adjustment time, adjustment reason, air compressor numbers involved, and specific adjustment parameters. Generate an operation report to summarize the air pressure regulation situation over a period of time, such as adjustment frequency, average air pressure stability, etc., to provide data support for subsequent system optimization and maintenance.

[0100] Furthermore, the air pressure regulation module uses a distributed collaborative control architecture to coordinate the control of multiple air compressor stations.

[0101] In a multi-unit oil-free air compressor station air pressure regulation system, the air pressure regulation module uses a distributed collaborative control architecture to coordinate the control of multiple air compressor stations. Under this architecture, each air compressor station is equipped with a local controller. The local controller is responsible for collecting data such as air pressure, gas production flow rate, and equipment operating status of the air compressor station where it is located, and realizes local data processing. The local controllers are connected through a high-speed communication network for real-time information interaction.

[0102] When the gas consumption demand of a certain air compressor station changes, such as an increase in demand causing a drop in air pressure, on the one hand, the local controller of this air compressor station adjusts the operating parameters of the air compressors in this station, such as changing the speed to regulate gas production; on the other hand, it transmits the information about the change in gas consumption demand to the local controllers of other air compressor stations. After receiving the information, the other local controllers adjust the operation of the air compressors according to the situation of their own air compressor stations. For example, the air compressor stations with sufficient gas production capacity increase gas production to jointly meet the overall gas consumption demand and maintain the stability of the system air pressure.

[0103] The distributed collaborative control architecture brings many advantages. In terms of response speed, the local controller can quickly respond to local changes, reduce the time from perception to response, and reduce air pressure fluctuations. In terms of reliability, there is no single point of failure. When some local controllers or communication links fail, other parts can still regulate to prevent the system from crashing. In terms of system expansion and maintenance, when adding a new air compressor station or renovating an existing air compressor station, the new local controller can be integrated by accessing the communication network. The maintenance of a single air compressor station does not affect other parts, reducing the complexity and cost of maintenance.

[0104] The above is only a preferred embodiment of the present invention, and does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the equivalent embodiments of equivalent changes within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. An air pressure control system for multiple oil-free air compressor stations, characterized in that: It includes a work monitoring module, a gas consumption estimation module and a gas pressure control module, and the work monitoring module, the gas consumption estimation module and the gas pressure control module are communicatively connected, wherein: The work monitoring module is used to monitor the current air pressure, gas production flow, operating status of gas-using equipment and gas flow of multiple air compressor stations; The gas consumption estimation module is used to establish a deep learning model of the operating status of gas-consuming equipment and gas consumption flow, and predict the gas consumption flow in the next time period based on the current operating status of the gas-consuming equipment and the gas consumption flow; The air pressure control module is used to adjust the number of operating air compressors and the air flow rate according to the air flow rate in the next time period to maintain a preset air pressure range.

2. The air pressure control system for multiple oil-free air compressor stations according to claim 1, characterized in that: The operating status of the gas-consuming equipment includes start-stop status, load status, working pressure, working flow and fault status.

3. The air pressure control system for multiple oil-free air compressor stations according to claim 1, characterized in that: The deep learning model includes a time series model and a random forest model. The time series model is used to predict the operating status of the gas-using equipment and the first gas flow in the next time period based on the current operating status of the gas-using equipment and the gas flow; the random forest model is used to predict the second gas flow in the next time period based on the operating status of the gas-using equipment in the next time period.

4. The air pressure control system for multiple oil-free air compressor stations according to claim 3, characterized in that: The gas flow rate in the next time period is obtained by comprehensive calculation of the first gas flow rate and the second gas flow rate.

5. The air pressure control system for multiple oil-free air compressor stations according to claim 4, characterized in that: The calculation formula is obtained by comprehensively calculating the first gas flow rate and the second gas flow rate: Q Z =w1Q1+w2Q2, In the formula, Q Z is the gas flow rate in the next time period; Q1 is the first gas flow rate, w1 is the corresponding first weight; Q2 is the second gas flow rate, w2 is the corresponding second weight; the first weight and the second weight are determined by the average error of the model's prediction in the historical period.

6. The air pressure control system for multiple oil-free air compressor stations according to claim 3, characterized in that: The time series model is configured as a long short-term memory network model, including the following construction steps: Data collection and integration: Collect historical data of multiple air compressor stations covering the operating status and gas flow of gas-consuming equipment from the work monitoring module, and process missing values ​​and abnormal values; Data preprocessing: normalize the data, organize it into a specific format in chronological order, and determine the time step to construct a time series dataset; Model architecture design: Determine the number of input layer neurons based on data characteristics, select an appropriate number of LSTM units to construct the LSTM layer, and set the number of output layer neurons based on the prediction target; Model compilation: Select mean square error as the loss function to measure the error, and Adam as the optimizer to update model parameters and set the learning rate; Model training: Divide the model into training set, validation set and test set. The training set is used to train the model. The number of training rounds and batch size are set for training. The validation set is used for model verification and fit optimization. Model evaluation and tuning: Use the test set to evaluate the model prediction performance, and optimize the model by adjusting hyperparameters and adding data samples.

7. The air pressure control system for multiple oil-free air compressor stations according to claim 3, characterized in that: The random forest model includes the following construction steps: Data preparation: Collect data including the operating status of gas-consuming equipment and the corresponding gas flow rate from the working monitoring module, and select key features that have a significant impact on gas flow prediction through correlation analysis method; Dataset division: Divide the preprocessed data into training set, validation set and test set; Model initialization: Determine the number of decision trees in the random forest model, set the number of randomly selected features, maximum depth, and minimum number of sample splits for each tree; Model training: Using the training set data, a unique training sample is constructed for each decision tree through the bootstrap sampling method with replacement; each tree randomly selects some features when dividing the node, and trains independently. Finally, the prediction results of all decision trees are combined to complete the training process of the random forest model; Model evaluation: First use the validation set to preliminarily evaluate the trained model and adjust the hyperparameters to optimize the model; The test set is then used for final evaluation to measure the generalization ability and prediction accuracy of the model on unknown data.

8. The air pressure control system for multiple oil-free air compressor stations according to claim 1, characterized in that: The method of adjusting the number of air compressors in operation and the air flow rate according to the air flow rate in the next time period to maintain a preset air pressure range includes the following steps: Gas flow analysis: Obtain the gas flow forecast data for the next time period from the gas consumption estimation module, combine it with the current air compressor's gas production capacity and the system's real-time gas pressure status, and evaluate the system's supply and demand status; Adjustment decision on the number of air compressors in operation: pre-set the threshold of the number of air compressors required to be operated based on different gas production scales, compare the results of gas flow analysis with the preset threshold, and increase the number of air compressors in operation if the predicted gas flow increases significantly and the current air compressor gas production capacity is difficult to match; conversely, if the predicted gas flow decreases and the current gas production capacity is obviously in excess, reduce the number of air compressors in operation; Start and stop decisions for specific air compressors: Develop a priority list for air compressors, and determine the priority of each air compressor based on equipment performance, service life, and maintenance records; adjust decisions based on the number of units in operation, and determine the specific air compressors that need to be started or stopped based on the priority list; Adjustment of air compressor operating parameters: For air compressors that are running or about to start, adjust the operating parameters according to the predicted air flow rate. The correspondence between the air compressor gas flow rate and the motor speed is preset. The required motor speed is calculated according to the predicted air flow rate, and then a command is sent to the inverter for adjustment. Real-time monitoring and feedback adjustment: After completing the number of air compressors in operation and parameter adjustment, conduct real-time and high-frequency monitoring of the actual air pressure and air flow, and compare the real-time data with the preset air pressure range. If the actual air pressure exceeds or approaches the preset range boundary, fine-tune the operating parameters or number of air compressors again according to the degree of deviation; Records and reports: Record each adjustment operation on the number of air compressors in operation and air flow, including the adjustment time, adjustment reason, air compressor number involved and specific adjustment parameters.

9. The air pressure control system for multiple oil-free air compressor stations according to claim 1, characterized in that: The air pressure control module adopts a distributed collaborative control architecture to coordinate the control of multiple air compressor stations.

Images