Internet of Things monitoring method for air separation device based on computer serial port

Abstract

The present invention relates to an Internet of Things monitoring method for an air separation device based on a computer serial port, and relates to the field of Internet of Things communication technology, including: monitoring and collecting air separation operation data in the air separation device through an air separation device operation station, performing local abnormality analysis, and obtaining local abnormality parameters; performing air separation abnormality discrimination according to local abnormality parameters, if not, transmitting the air separation operation data to an Internet of Things communication module through a computer serial port, and if so, directly transmitting to the Internet of Things communication module; when it is yes, sending the air separation operation data to a cloud server, configuring an abnormal continuous identification coefficient according to the data characteristics of the air separation operation data and the local abnormality parameters, performing abnormal continuous integrated identification, and obtaining an abnormal continuous window; configuring the air separation device operation station according to the abnormal continuous window. The present invention solves the technical problem of network risks when an air separation device is monitored by the Internet of Things in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things communication, and in particular to an Internet of Things monitoring method for an air separation device based on a computer serial port. Background Art

[0002] In the industrial production of air separation units, the Industrial Internet of Things is used to monitor equipment and production to ensure stable production. Conventional Industrial Internet of Things uses the external network to directly connect to the operator station of the air separation unit, or to the industrial control network. Even if the network is added with devices such as firewalls and isolation gateways, there are still security risks in the control network.

[0003] If the industrial Internet of Things is not connected, many functions of the industrial Internet of Things cannot be applied, such as production monitoring, resulting in low efficiency and difficulty in production monitoring. Therefore, the existing technology has the technical problems of poor monitoring effect of the industrial Internet of Things of air separation units and high network security risks. Summary of the invention

[0004] Aiming at the technical problems in the prior art of poor industrial Internet of Things monitoring effect of air separation devices and high network security risks, the present invention proposes an Internet of Things monitoring method for air separation devices based on computer serial ports.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] The present invention provides an air separation device IoT monitoring method based on a computer serial port, comprising: monitoring and collecting air separation operation data in the air separation device through an air separation device operation station, performing local abnormality analysis on the air separation operation data, and obtaining local abnormality parameters, wherein the local abnormality analysis includes flow abnormality analysis and access abnormality analysis;

[0007] Perform air separation abnormality judgment according to the local abnormal parameters, if not, transmit the air separation operation data to the Internet of Things communication module through the computer serial port, if yes, transmit the air separation operation data directly to the Internet of Things communication module;

[0008] When the air separation abnormality is judged as yes, the air separation operation data is sent to the cloud server, and according to the data characteristics of the air separation operation data and the local abnormality parameters, an abnormality continuous identification coefficient is configured, and an abnormality continuous integrated identification is performed to obtain an abnormality continuous window;

[0009] The air separation device operation station is configured according to the abnormality persistence window, and the air separation operation data is directly transmitted to the Internet of Things communication module within the abnormality persistence window after the current moment for Internet of Things monitoring.

[0010] The beneficial effects of the present invention are: by adding a one-way transmission mechanism to the computer serial port, the access of the external network to the industrial control network is blocked, the network security is significantly improved, and at the same time, combined with the local abnormality analysis of the air separation operation data (including flow abnormality analysis and access abnormality analysis), the accuracy and reliability before data transmission are guaranteed. With the support of the air separation abnormality discrimination mechanism, the data transmission strategy is dynamically adjusted according to the operating status. When there is no abnormality, the data is transmitted to the Internet of Things communication module through the computer serial port. When there is an abnormality, it is directly uploaded to the cloud server and the abnormality duration identification coefficient is configured. Through integrated machine learning, multi-path abnormality duration prediction is performed, and the abnormal duration window is accurately configured, so as to further improve the accuracy and real-time monitoring of air separation production abnormalities, ensure the stability of air separation production, and effectively improve the accuracy and response speed of air separation production abnormality monitoring. While ensuring the security of the industrial network of the air separation device, the present invention achieves the goals of high monitoring efficiency, timely response, and safe operation of the air separation device. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A flow diagram of the method for monitoring the Internet of Things of an air separation device based on a computer serial port provided by the present invention;

[0012] Figure 2 A schematic flow chart of local anomaly analysis in the Internet of Things monitoring method for an air separation device based on a computer serial port provided by the present invention. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0014] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0015] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0016] Embodiment 1, as Figure 1 As shown, an embodiment of the present invention provides an air separation device Internet of Things monitoring method based on a computer serial port. The method is applied to an air separation device Internet of Things monitoring system based on a computer serial port. The system includes an air separation device operation station and an Internet of Things communication module connected through a computer serial port, and a cloud server.

[0017] The IoT communication module connects to the air separation unit operation station through the IoT, monitors the operation data of the air separation unit in the air separation unit operation station, and uploads it to the cloud server for equipment operation status analysis, etc., to realize air separation production based on the IoT. The air separation unit is, for example, an industrial equipment for separating oxygen, nitrogen and argon in the air, and is used to produce separation gas.

[0018] The air separation unit operation station and the IoT communication module are connected through a computer serial port. For example, the data is transmitted to the IoT communication module in one direction through the computer serial port RS232 protocol, and the IoT communication module then transmits the data to a data center, such as a cloud server, through the Internet.

[0019] By transmitting data to the IoT communication module through the computer serial port in one direction, the industrial network of the air separation unit operation station can be effectively physically isolated from the Internet, which can protect the industrial control network of the air separation unit from virus attacks on the Internet and has higher security.

[0020] However, the transmission rate of the computer serial port RS232 protocol is low, and there is no data verification mechanism, so there may be data transmission anomalies. When production anomalies occur in the air separation unit, the transmission timeliness and quality may be low, and the efficient transmission of the production monitoring data of the air separation unit cannot be guaranteed. Therefore, this application sets the following steps to monitor whether the air separation operation data is abnormal and determine whether to enable the computer serial port to transmit data, so as to protect the industrial control network of the air separation unit operation station, or improve the timeliness and quality of monitoring data transmission.

[0021] Specifically, there are two data transmission channels between the air separation unit operation station and the IoT communication module. The first data transmission channel is based on one-way transmission of the computer serial port (RS232 protocol) and is used for data transmission during daily stable operation. This channel can physically isolate external network access to ensure the network security of the air separation unit operation station. The second data transmission channel is based on two-way transmission of network protocols (such as TCP / IP) and is used for rapid data transmission during abnormal operation.

[0022] The method specifically comprises the following steps:

[0023] S100: monitoring and collecting air separation operation data in the air separation device through an air separation device operation station, performing local abnormality analysis on the air separation operation data, and obtaining local abnormality parameters, wherein the local abnormality analysis includes flow abnormality analysis and access abnormality analysis.

[0024] In the embodiment of the present application, the air separation operation data in the air separation unit is monitored and collected in real time through the air separation unit operation station, and the data includes key operation indicators, such as gas flow data, etc., to obtain the air separation operation data.

[0025] Then, based on the air separation operation data, an abnormal analysis of the air separation unit operation status is performed locally at the air separation unit operation station to obtain local abnormal parameters. The abnormal analysis includes abnormal analysis in two dimensions, specifically, flow abnormal analysis based on gas flow and access abnormal analysis of external network access to the industrial network of the air separation unit operation station.

[0026] By performing abnormal analysis on the air separation operation data, it is determined whether the current operation of the air separation unit is abnormal, and whether to enable the computer serial port protocol to transmit the air separation operation data between the air separation unit operation station and the Internet of Things communication module, or directly transmit the air separation operation data through the network within the air separation unit operation station and the Internet of Things communication module.

[0027] Step S100 in the method provided in the embodiment of the present application includes:

[0028] The air separation operation data in the air separation device is monitored and collected through the air separation device operation station, wherein the air separation operation data includes gas flow data.

[0029] Perform flow anomaly analysis on the air separation operation data to obtain flow anomaly parameters.

[0030] An access anomaly analysis is performed on the air separation operation data to obtain access anomaly parameters.

[0031] A local abnormality parameter is obtained by calculation according to the traffic abnormality parameter and the access abnormality parameter.

[0032] In the embodiment of the present application, the air separation operation data in the air separation device is monitored and collected in real time through the air separation device operation station, including key gas flow data (such as the volume of oxygen gas passing through the air separation device per minute). Exemplarily, the air separation operation data is the gas flow data.

[0033] Then, the gas flow rate data in the air separation operation data is analyzed for gas flow rate anomaly, and the degree of anomaly of the gas flow rate data in the current production operation of the air separation unit is obtained as a flow rate anomaly parameter.

[0034] The step of "performing flow anomaly analysis on the air separation operation data to obtain flow anomaly parameters" in the method provided in the embodiment of the present application includes:

[0035] Extract the historical gas flow data sequence within the preset time zone past the current moment;

[0036] Randomly extracting a number of historical gas flow data from the historical gas flow data sequence, calculating the deviation from the gas flow data, and obtaining the flow deviation amplitude;

[0037] The flow deviation amplitude is used as a flow anomaly parameter.

[0038] In the embodiment of the present application, historical gas flow data in a preset time zone past the current moment is extracted to obtain a historical gas flow data sequence. The preset time zone refers to a time window for analysis, such as all monitored and recorded gas flow data within 10 minutes or 1 hour past the current moment.

[0039] Randomly extract a number of historical gas flow data from the above historical gas flow data sequence. For example, randomly extract 20 historical gas flow data. Random extraction can avoid the deviation caused by selecting sequential data and ensure the comprehensiveness of the analysis.

[0040] Furthermore, the deviation between several historical gas flow data and the current gas flow data is calculated. The deviation can reflect the abnormality of the current production and operation of the air separation unit. The larger the deviation, the greater the current abnormality.

[0041] Exemplarily, the absolute value of the difference between the current gas flow data and the mean of several historical gas flow data is calculated, and the ratio to the current gas flow data is used as the flow deviation amplitude. Exemplarily, the current flow value is 1000 Nm³ / h, and the mean of several historical gas flow data is 950 Nm³ / h, then the deviation amplitude is 5%. The smaller the current flow deviation amplitude is, the smaller the difference between the current gas flow data and the gas flow in the production process of the air separation unit in the historical time is, and the smaller the abnormality of the current production is.

[0042] Furthermore, the flow deviation amplitude is directly used as a flow anomaly parameter.

[0043] Through the above steps, the abnormal fluctuation of air separation gas flow can be analyzed in real time, and the abnormal degree of the operation status of the air separation equipment can be dynamically evaluated. In this way, possible equipment failures or pipeline problems can be obtained, thereby ensuring the safety and stability of the operation of the air separation unit.

[0044] Furthermore, the air separation operation data is analyzed for abnormalities in the industrial network of the air separation unit operation station accessed by the external network, and access abnormality parameters are obtained to reflect the abnormalities in the network dimension of the current air separation unit operation station.

[0045] The step of "performing access anomaly analysis on the air separation operation data to obtain access anomaly parameters" in the method provided in the embodiment of the present application includes:

[0046] The timestamp of the air separation unit operation station being accessed by the external network in the past time from the current moment is obtained, and the maximum abnormal access time period is extracted.

[0047] The real-time abnormal access time period is calculated based on the timestamp of the last time the air separation unit operation station was accessed by the external network.

[0048] The ratio of the real-time abnormal access time period to the maximum abnormal access time period is calculated, and an access abnormality parameter is obtained by calculation.

[0049] In the embodiment of the present application, the timestamps of all external network accesses in the past period of time (such as the last month) are extracted from the network access log of the air separation unit operation station. The timestamp is an identifier that accurately records the time when the access event occurred, for example, in the format of 2025-01-14 10:30:45. These timestamps constitute the time series of access behaviors and provide a data basis for subsequent analysis. There can be multiple timestamps of all external network accesses in the past period of time, forming an external access timestamp sequence.

[0050] Furthermore, according to the above external access timestamp sequence, the intervals between two adjacent access timestamps are calculated to obtain the access cycle sequence, and then the maximum access cycle is extracted as the maximum abnormal access time cycle. Among them, the external network access to the air separation unit operation station may involve risky behaviors such as attacking the industrial network, so the maximum access cycle is taken as the maximum abnormal access time cycle.

[0051] Furthermore, the timestamp of the last external network access before the current moment is obtained, and combined with the timestamp corresponding to the air separation operation data at the current moment, the real-time abnormal access time period if the external network access currently occurs is calculated. Among them, external network access may pose a risk to the industrial network (which can be protected by firewalls, etc.). The longer the real-time abnormal access time period, the longer the accumulated time length without external network access, the higher the probability of external network access, and the greater the network risk of the air separation unit operation station, which requires physical isolation protection. Conversely, the smaller the real-time abnormal access time period, the smaller the network risk of the air separation unit operation station, and the air separation operation data can be directly transmitted to the IoT communication module for analysis, which can also ensure the security of the industrial network of the air separation unit operation station.

[0052] Specifically, the ratio of the real-time abnormal access time period to the maximum abnormal access time period is calculated, and then 1 minus the ratio is used as the access abnormality parameter. The smaller the real-time abnormal access time period, the larger the access abnormality parameter, and the more the air separation operation data can be directly transmitted to the Internet of Things communication module, so as to perform more efficient and accurate analysis of the operation of the air separation unit.

[0053] Through the above analysis steps, we can quickly capture abnormal changes in network access frequency, analyze the network security of the air separation unit operation station, respond to possible external threats in a timely manner, and ensure the safe and stable operation of the industrial control network under the premise of analyzing the operation of the air separation unit.

[0054] Further, according to the flow abnormality parameter and the access abnormality parameter, the local abnormality parameter is calculated. For example, the flow abnormality parameter and the access abnormality parameter are weighted to obtain the local abnormality parameter. In the weighted calculation, since the operation stability of the air separation unit is more important, the weight of the flow abnormality parameter is greater than the access abnormality parameter, for example, the weights of the flow abnormality parameter and the access abnormality parameter are 0.7 and 0.3.

[0055] By analyzing the volatility of the air separation flow of the air separation unit and the changes in the external network access cycle, the abnormal operation of the current air separation unit operation station and the degree of network risk can be analyzed. When the gas flow fluctuates greatly and the external network access cycle is small (the network risk is small), the air separation operation data can be directly transmitted to the IoT communication module to analyze the abnormal operation of the air separation unit to ensure the production and operation stability of the air separation unit. The above steps are performed locally at the air separation unit operation station. The above steps are all direct calculation steps, which consume less computing power, do not require external computing power support, have high processing efficiency, and can quickly analyze and obtain the local abnormal parameters of the current air separation unit operation to reflect the abnormal operation.

[0056] S200: Performing air separation abnormality determination according to the local abnormality parameters, if not, transmitting the air separation operation data to the IoT communication module through the computer serial port, if yes, directly transmitting the air separation operation data to the IoT communication module;

[0057] In the embodiment of the present application, the air separation abnormality is judged according to the local abnormal parameter, for example, whether the local abnormal parameter is greater than or equal to the local abnormal parameter threshold value is judged, and the local abnormal parameter threshold value is set according to the amplitude of the abnormal fluctuation of the gas flow rate during the operation of the air separation unit and the size of the external access cycle, and is set based on the technical personnel in this field, for example, to 0.37. In this way, even if the flow abnormality parameter is 1, the local abnormality parameter is only 0.3 when the weight is 0.3, and there will be no misjudgment, and when the gas flow fluctuation reaches 10%, it can be judged that the air separation unit is abnormal, and the air separation operation data is directly transmitted to the Internet of Things communication module through the second data transmission channel, so as to perform efficient and high-quality abnormal analysis of the air separation unit. The local abnormal parameter threshold value can be adjusted by the technical personnel in this field according to the actual production and operation conditions, and can also be 0.2.

[0058] At the same time, even if the gas flow rate fluctuates abnormally, if the access abnormal parameters are small, the probability of network risk in the air separation unit operation station is high, and the local abnormal parameters are small, the industrial network can be prevented from being invaded by the network. However, if the gas flow rate fluctuates greatly and the flow abnormal parameters are large, in order to ensure the stable operation of the air separation unit, the air separation operation data can still be directly transmitted to the IoT communication module through the second data transmission channel to ensure the stability of the device operation.

[0059] Determine whether the local abnormal parameter is greater than or equal to the local abnormal parameter threshold. If not, the air separation unit is currently operating normally, and the air separation operation data is transmitted to the Internet of Things communication module through the first data transmission channel of the computer serial port to ensure the network security of the air separation unit operation station. If so, the air separation operation data is directly transmitted to the Internet of Things communication module through the second data transmission channel, and high-computing-power machine learning and other means are used through the external network to analyze the operation status of the air separation unit and make adjustments.

[0060] S300: When the air separation abnormality is judged to be yes, the air separation operation data is sent to the cloud server, and according to the data characteristics of the air separation operation data and the local abnormality parameters, an abnormality continuous identification coefficient is configured, and an abnormality continuous integrated identification is performed to obtain an abnormality continuous window;

[0061] In the embodiment of the present application, when an abnormal operation occurs in the air separation unit, the abnormality may occur in a subsequent period of time, for example, the abnormality of gas flow will last for a period of time. Therefore, when the air separation abnormality is judged to be yes, the air separation operation data is sent to the cloud server, and the abnormality duration identification coefficient is configured according to the data characteristics of the air separation operation data and the local abnormality parameters, and the abnormality duration is integrated to obtain the abnormality duration window. During the time of the abnormal duration window after the current moment, the data is directly transmitted to the Internet of Things communication module through the second data transmission channel to perform abnormal analysis and adjustment of the operation of the air separation unit to maintain the stable operation of the air separation unit.

[0062] Among them, the identification configuration of the abnormal persistence window needs to be relatively accurate. The embodiment of the present application uses integrated machine learning for analysis to improve its accuracy, avoid the abnormal persistence window being too large, causing external networks to attack the network of the air separation unit operation station, and avoid the abnormal persistence window being too small, causing the air separation unit to remain in an abnormal state and stop the transmission of high-speed and high-quality air separation operation data, resulting in a decrease in the accuracy and efficiency of the air separation unit adjustment analysis. Since the identification configuration of the abnormal persistence window requires a large computing power, it is deployed on a cloud server for identification configuration to improve processing efficiency and accuracy.

[0063] Step S300 in the method provided in the embodiment of the present application includes:

[0064] When the air separation abnormality is judged to be yes, the air separation operation data is sent to the cloud server.

[0065] The maximum gas flow rate during the operation of the air separation device is obtained, and the ratio of the gas flow rate data in the air separation operation data to the maximum gas flow rate is calculated to obtain a first abnormality continuous identification coefficient.

[0066] The local abnormality parameter is used as a second abnormality persistence identification coefficient.

[0067] A configuration abnormality persistence identification coefficient is calculated according to the first abnormality persistence identification coefficient and the second abnormality persistence identification coefficient.

[0068] According to the abnormal persistence identification coefficient, abnormal persistence integrated identification is performed to obtain the abnormal persistence window.

[0069] In the embodiment of the present application, when the air separation abnormality is judged to be yes, the current air separation operation data is uploaded to the cloud server, and the abnormal continuous window for continuously transmitting the air separation operation data through the second data transmission channel is configured.

[0070] According to the data characteristics of the air separation operation data and the local abnormality parameters, the abnormality persistence identification coefficient is configured, and the subsequent integrated identification configuration of the abnormality persistence window is performed.

[0071] Specifically, the maximum gas flow rate during the operation of the air separation unit is obtained, and the ratio of the gas flow rate data in the current air separation operation data to the maximum gas flow rate is calculated to obtain the first abnormality continuous identification coefficient. The maximum gas flow rate is the maximum gas flow rate during the operation in the historical operation data of the air separation unit, such as the maximum oxygen flow rate.

[0072] The larger the gas flow data in the current air separation operation data, the larger the scale of the current air separation production, and the greater the impact of an abnormality. Therefore, it is necessary to improve the accuracy of the abnormal continuous window recognition configuration, and the larger the first abnormal continuous recognition coefficient.

[0073] Furthermore, the local abnormality parameter is used as the second abnormality continuity identification coefficient. The greater the local abnormality level of the current air separation unit operating station, the greater the gas flow abnormality amplitude, and the accuracy of the abnormal continuity window identification configuration is improved to continuously monitor possible gas flow fluctuations and control and adjust the air separation unit to stabilize air separation production.

[0074] According to the first abnormality persistence identification coefficient and the second abnormality persistence identification coefficient, the abnormality persistence identification coefficient is calculated, for example, the average of the first abnormality persistence identification coefficient and the second abnormality persistence identification coefficient is calculated as the abnormality persistence identification coefficient. The first abnormality persistence identification coefficient and the second abnormality persistence identification coefficient are both natural numbers between 0 and 1, and the abnormality persistence identification coefficient is also a natural number falling within the range of 0 to 1.

[0075] According to the abnormal persistence recognition coefficient, the abnormal persistence integrated recognition is performed to obtain the abnormal persistence window. The abnormal persistence recognition coefficient reflects the accuracy of the abnormal persistence window integrated recognition.

[0076] Step S300 in the method provided in the embodiment of the present application also includes:

[0077] Integrated machine learning is adopted to train a space-separation anomaly duration predictor including K space-separation anomaly duration prediction paths, the output feature of each space-separation anomaly duration prediction path is the space-separation anomaly duration, and K is an integer greater than 1.

[0078] According to the abnormality duration identification coefficient and K, the number of predicted paths Y for predicting the duration of the air separation abnormality on the air separation operation data is calculated, where Y is an integer greater than or equal to 1 and less than or equal to K.

[0079] Y space-separation anomaly duration prediction paths are randomly selected, the space-separation operation data is input, and the space-separation anomaly durations of Y paths are obtained through prediction output.

[0080] According to the Y path space-divided anomaly durations, the anomaly duration window is calculated.

[0081] In an embodiment of the present application, integrated machine learning is used to identify and predict the duration of an abnormality in an air separation unit based on air separation operation data, that is, the duration of an air separation abnormality when gas flow data is abnormal, so as to improve the accuracy of identification and prediction.

[0082] Among them, integrated machine learning is used to train an air separation anomaly duration predictor including K air separation anomaly duration prediction paths, the input feature of each air separation anomaly duration prediction path is the air separation operation data (i.e., gas flow data) with abnormal fluctuations, and the output feature is the air separation anomaly duration, and K is an integer greater than 1, for example, 10.

[0083] The step of "using integrated machine learning to train an air-separation anomaly duration predictor including K air-separation anomaly duration prediction paths" in the method provided in the embodiment of the present application includes:

[0084] According to the abnormal operation record log of the air separation unit, a sample air separation operation data set is collected as an input feature, and the air separation abnormality duration after different sample air separation operation data is collected to obtain a sample air separation abnormality duration set as an output feature;

[0085] Combining the sample space separation operation data set and the sample space separation anomaly duration set to obtain a space separation anomaly duration training sample set;

[0086] Randomly extracting a preset proportion of training samples from the space-separation anomaly duration training sample set with replacement, and using machine learning to train and obtain a first space-separation anomaly duration prediction path;

[0087] Continue to randomly extract training samples, train K space-separation anomaly duration prediction paths, and obtain the space-separation anomaly duration predictor.

[0088] In an embodiment of the present application, according to the abnormal operation record log of the air separation device, sample air separation operation data when an abnormality occurs, that is, air separation operation data when gas flow data fluctuates and the local abnormal parameter is greater than the local abnormal parameter threshold, is collected as an input feature for predicting the duration of air separation abnormality. Also, the length of time that the air separation operation data continues to be abnormal after each sample air separation operation data is collected, such as the duration when the gas flow data fluctuation reaches 10%, or the duration when the local abnormal parameter is greater than the local abnormal parameter threshold, is used as the duration of air separation abnormality, as an output feature for predicting the duration of air separation abnormality. In this way, a set of sample air separation abnormality durations is obtained. Exemplarily, the duration of sample air separation abnormality is 30 minutes.

[0089] The above-mentioned input features are combined with the output features, that is, the sample space-separation operation data set and the sample space-separation anomaly duration set are combined to obtain the space-separation anomaly duration training sample set for training the anomaly duration predictor.

[0090] In the space-separation anomaly duration training sample set, a preset proportion of samples (such as 70%) are randomly selected by sampling with replacement as the training data for training the first space-separation anomaly duration prediction path based on integrated machine learning, that is, the first batch of training data. The first batch of training data includes multiple sample space-separation operation data and corresponding multiple sample space-separation anomaly durations.

[0091] Furthermore, based on machine learning, the first batch of training data is used to train the first space-separation anomaly duration prediction path.

[0092] Exemplarily, a feedforward neural network in machine learning is used to train the first air separation abnormality duration prediction path. Exemplarily, the first batch of training data is normalized first, and then divided into a training set and a validation set in a ratio of 8:2. A feedforward neural network is used to construct the first air separation abnormality duration prediction path, which includes an input layer, an output layer and multiple hidden layers. The number of input nodes in the input layer is 1, corresponding to the gas flow in the air separation operation data, and the number of output nodes in the output layer is 1, corresponding to the air separation abnormality duration. Each hidden layer includes multiple neuron nodes, and a ReLU activation function is used. The loss function in the training process adopts the mean square error function (MSE), and then the training set is used for training. According to the calculated loss, the gradient descent optimization algorithm (such as Adam) is used to adjust the network parameters, minimize the loss function, and improve the prediction accuracy. Iterative training is performed in this way until the training converges, for example, the loss is less than or equal to 5%, and the training is completed. After the training is completed, the validation set is used for accuracy test verification. If the accuracy meets the requirements, such as greater than or equal to 95%, the verification is passed and the training is completed. Otherwise, the training is continued until it meets the requirements.

[0093] After the first space-separation anomaly duration prediction path is completed, based on the same steps, continue to randomly extract a preset proportion of training samples in the space-separation anomaly duration training sample set, use machine learning to train other K-1 space-separation anomaly duration prediction paths, and combine the K space-separation anomaly duration prediction paths that have been trained to obtain the space-separation anomaly duration predictor.

[0094] Among them, since the training data of the K space-separation anomaly duration prediction paths are randomly extracted with replacement, the training data are not exactly the same. We can train to obtain multiple space-separation anomaly duration prediction paths with different performances, perform space-separation anomaly duration predictions separately, and then fuse the prediction results of multiple paths to improve the prediction accuracy.

[0095] Furthermore, according to the current abnormality duration identification coefficient and K, the number of predicted paths Y for predicting the duration of the air separation abnormality for the air separation operation data is calculated, where Y is an integer greater than or equal to 1 and less than or equal to K.

[0096] For example, the larger the abnormality duration identification coefficient is, the greater the abnormality of the current air separation operation data is and the larger the air separation production scale is, the more accurate the air separation abnormality duration identification prediction is needed to improve the accuracy and reliability of abnormality monitoring. For example, the abnormality duration identification coefficient is multiplied by K and rounded up to obtain Y, which is greater than or equal to 1 and less than or equal to K.

[0097] According to Y, Y space-division anomaly duration prediction paths are randomly selected from the K space-division anomaly duration prediction paths, the current space-division operation data is input into the Y space-division anomaly duration prediction paths, and the prediction outputs are respectively identified to obtain the space-division anomaly durations of the Y paths. The space-division anomaly durations of the Y paths may be different or the same.

[0098] According to the spatial anomaly duration of the Y paths, the anomaly duration window is calculated, so that the output result of the Y spatial anomaly duration prediction paths is fused with higher accuracy.

[0099] The step of "calculating the abnormal duration window according to the Y path space-division abnormal durations" in the method provided in the embodiment of the present application includes:

[0100] The test obtains the path accuracy of the Y space-division anomaly duration prediction paths, and weightedly calculates the space-division anomaly durations of the Y paths to obtain the space-division anomaly duration.

[0101] A time window greater than or equal to the duration of the space-division anomaly is configured as the anomaly duration window.

[0102] In an embodiment of the present application, the accuracy of K space-division anomaly duration prediction paths is tested and calculated based on the validation set in the training process to obtain K path accuracy rates as the accuracy rates of the K space-division anomaly duration prediction paths.

[0103] Then, the Y path accuracy rates of the currently selected Y space-division anomaly duration prediction paths are used to weightedly calculate and fuse the space-division anomaly durations of the Y paths to obtain the space-division anomaly duration.

[0104] Exemplarily, the ratio of each path accuracy to the sum of Y path accuracy is used as a weight to obtain Y weights, and then the Y path space-division anomaly durations are weighted and fused to obtain the space-division anomaly duration. Weighted calculation fusion based on accuracy can further improve the accuracy of the space-division anomaly duration. The higher the accuracy of the space-division anomaly duration prediction path, the greater the proportion of the space-division anomaly duration of the path in the final space-division anomaly duration.

[0105] Furthermore, a time window greater than or equal to the duration of the space-separation anomaly is configured as the anomaly duration window. For example, the duration of the space-separation anomaly is directly used as the time window, or the duration of the space-separation anomaly is rounded up to the integer of 1 minute or 5 minutes as the time window. For example, if the duration of the space-separation anomaly is 31 minutes, the rounded time window is 35 minutes to obtain the anomaly duration window.

[0106] S400: The air separation device operation station is configured according to the abnormality persistence window, and the air separation operation data is directly transmitted to the Internet of Things communication module within the abnormality persistence window after the current moment to perform Internet of Things monitoring.

[0107] In the embodiment of the present application, the abnormal duration window is a time period calculated by abnormal duration identification, which is used to indicate the possible duration of the current abnormal state. For example, if the duration of the current abnormal state is predicted to be 30 minutes, the abnormal duration window is from the current moment to 30 minutes later.

[0108] When the air separation abnormality is judged as yes and the second data transmission channel is opened, the abnormal duration window is transmitted to the air separation unit operation station through the Internet of Things communication module. Within the abnormal duration window after the current moment, the air separation operation data of the air separation unit operation station is directly transmitted to the Internet of Things communication module to monitor the abnormal operation of the air separation unit, ensure the efficiency and quality of data transmission, so as to monitor and adjust the abnormal production of the air separation unit and improve the reliability of air separation production until the abnormal duration window ends. After the abnormal duration window ends, the first data transmission channel can be restored, and one-way transmission can be performed through the computer serial port R232 protocol to ensure industrial network security. When transmitting through the second data transmission channel, firewalls and other measures can be used to ensure network security.

[0109] The method for monitoring the Internet of Things of an air separation device based on a computer serial port provided by the embodiment of the present invention has at least the following technical effects:

[0110] By introducing a dynamic configuration mechanism and a dual-channel transmission architecture, an efficient and flexible monitoring and response mechanism is established between the air separation unit operation station and the IoT communication module. The computer serial port (RS232 protocol) is used as a one-way transmission channel during stable operation to physically isolate the access risk of the external network to the operation station, which fundamentally improves network security. In the event of abnormal operation, real-time transmission of key data is achieved through an efficient network channel to ensure the efficiency, security and integrity of data transmission. Through local abnormality analysis and cloud-based continuous integrated identification of abnormalities, the duration of abnormalities can be accurately predicted, and the configuration of the operation station can be dynamically adjusted within the abnormal duration window, which improves the quality and efficiency of data transmission during abnormal operation and significantly improves the accuracy of monitoring and the timeliness of response. After the abnormality ends, it automatically recovers to the computer serial port transmission mode to ensure the rational use of transmission resources and the smooth transition of the operating status. The overall solution achieves an effective balance between network security, data transmission efficiency and abnormal monitoring accuracy, while reducing operation and maintenance costs, significantly improving the safety and operation stability of the air separation unit, and providing strong support for efficient industrial IoT monitoring.

[0111] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0112] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] The present invention is described with reference to the flow diagrams and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flow diagram and / or block diagram, as well as the combination of flows and / or blocks in the flow diagram and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the flow diagram and / or block diagram. Figure 1 flow or flows and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0114] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the flow Figure 1 flow or flows and / or boxes Figure 1 A function specified in one or more boxes.

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the flow Figure 1 flow or flows and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0116] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0117] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. An air separation device IoT monitoring method based on a computer serial port, characterized in that: The method is applied to an air separation device IoT monitoring system based on a computer serial port, the system comprising an air separation device operation station and an IoT communication module connected via a computer serial port, and a cloud server, the method comprising: Monitoring and collecting air separation operation data in the air separation device through the air separation device operation station, performing local abnormality analysis on the air separation operation data, and obtaining local abnormality parameters, wherein the local abnormality analysis includes flow abnormality analysis and access abnormality analysis; Perform air separation abnormality judgment according to the local abnormal parameters, if not, transmit the air separation operation data to the Internet of Things communication module through the computer serial port, if yes, transmit the air separation operation data directly to the Internet of Things communication module; When the air separation abnormality is judged as yes, the air separation operation data is sent to the cloud server, and according to the data characteristics of the air separation operation data and the local abnormality parameters, an abnormality continuous identification coefficient is configured, and an abnormality continuous integrated identification is performed to obtain an abnormality continuous window, including: Acquire a maximum gas flow rate during the operation of the air separation device, calculate a ratio of the gas flow rate data in the air separation operation data to the maximum gas flow rate, and obtain a first abnormality continuous identification coefficient; Using the local abnormality parameter as a second abnormality persistence identification coefficient; Calculate a configuration abnormality continuous identification coefficient according to the first abnormality continuous identification coefficient and the second abnormality continuous identification coefficient; According to the abnormality persistence identification coefficient, abnormality persistence integrated identification is performed to obtain the abnormality persistence window, including: By adopting integrated machine learning, a space-separation anomaly duration predictor including K space-separation anomaly duration prediction paths is trained, the output feature of each space-separation anomaly duration prediction path is the space-separation anomaly duration, and K is an integer greater than 1; According to the abnormality duration identification coefficient and K, the number of predicted paths Y for predicting the duration of the air separation abnormality for the air separation operation data is calculated, where Y is an integer greater than or equal to 1 and less than or equal to K; Randomly select Y air separation anomaly duration prediction paths, input the air separation operation data, and predict and output the air separation anomaly durations of the Y paths; According to the Y path space-division anomaly durations, an anomaly duration window is calculated; The air separation device operation station is configured according to the abnormality persistence window, and the air separation operation data is directly transmitted to the Internet of Things communication module within the abnormality persistence window after the current moment for Internet of Things monitoring.

2. The method for monitoring the air separation device based on computer serial port according to claim 1, characterized in that: The air separation operation data in the air separation unit is monitored and collected through the air separation unit operation station, and local abnormality analysis is performed on the air separation operation data to obtain local abnormality parameters, including: Monitoring and collecting air separation operation data in the air separation device through an air separation device operation station, wherein the air separation operation data includes gas flow data; Performing flow anomaly analysis on the air separation operation data to obtain flow anomaly parameters; Performing access anomaly analysis on the air separation operation data to obtain access anomaly parameters; A local abnormality parameter is obtained by calculation according to the traffic abnormality parameter and the access abnormality parameter.

3. The method for monitoring the air separation device based on the computer serial port according to claim 2 is characterized in that: Performing flow anomaly analysis on the air separation operation data to obtain flow anomaly parameters includes: Extract the historical gas flow data sequence within the preset time zone past the current moment; Randomly extracting a number of historical gas flow data from the historical gas flow data sequence, calculating the deviation from the gas flow data, and obtaining the flow deviation amplitude; The flow deviation amplitude is used as a flow anomaly parameter.

4. The method for monitoring the air separation device based on the computer serial port according to claim 2 is characterized in that: Performing access anomaly analysis on the air separation operation data to obtain access anomaly parameters includes: Obtain the timestamp of the air separation unit operation station being accessed by the external network in the past time from the current moment, and extract the maximum abnormal access time period; Calculate and obtain a real-time abnormal access time period according to the timestamp of the last time the air separation unit operation station was accessed by an external network; The ratio of the real-time abnormal access time period to the maximum abnormal access time period is calculated, and an access abnormality parameter is obtained by calculation.

5. The method for monitoring the air separation device based on computer serial port according to claim 1, characterized in that: Integrated machine learning is used to train an air-space anomaly duration predictor including K air-space anomaly duration prediction paths, including: According to the abnormal operation record log of the air separation unit, a sample air separation operation data set is collected as an input feature, and the air separation abnormality duration after different sample air separation operation data is collected to obtain a sample air separation abnormality duration set as an output feature; Combining the sample space separation operation data set and the sample space separation anomaly duration set to obtain a space separation anomaly duration training sample set; Randomly extracting a preset proportion of training samples from the space-separation anomaly duration training sample set with replacement, and using machine learning to train and obtain a first space-separation anomaly duration prediction path; Continue to randomly extract training samples, train K space-separation anomaly duration prediction paths, and obtain the space-separation anomaly duration predictor.

6. The method for monitoring the air separation device based on computer serial port according to claim 1, characterized in that: According to the Y path space-division anomaly durations, the anomaly duration window is calculated, including: The path accuracy of the Y space-division anomaly duration prediction paths is obtained by testing, and the space-division anomaly duration of the Y paths is weightedly calculated to obtain the space-division anomaly duration; A time window greater than or equal to the duration of the space-division anomaly is configured as the anomaly duration window.

Images