Online monitoring method, device and system for ethanol production concentration

Abstract

The invention relates to the technical field of ethanol production, and discloses an ethanol production concentration online monitoring method, device and system, and the method comprises the steps: collecting historical data in an ethanol production process, and carrying out the preprocessing to form a feature data set; based on the data set, an ethanol concentration prediction model is constructed, and a deep learning algorithm is utilized to predict a future concentration change trend. And obtaining production state data and environmental condition data in real time, and inputting the data into the prediction model to obtain a predicted concentration value. Detecting the concentration value in real time by combining a sensor on the ethanol production line, comparing and analyzing the concentration values, judging whether deviation exists or not according to a preset concentration deviation threshold value, and recording a deviation point. According to the number of the deviation points, the deviation value and a preset evaluation rule, the concentration control state is evaluated, if an abnormity is found, an alarm mechanism is triggered, and an evaluation result and alarm information are sent to related personnel in time. The intelligent level of concentration monitoring in the ethanol production process is improved, and potential problems can be found and prevented in advance.

Description

Technical Field

[0001] The invention relates to the technical field of ethanol production, and in particular to an online monitoring method, device and system for ethanol production concentration. Background Art

[0002] In the ethanol production process, the monitoring and control of concentration is a crucial link, which is directly related to the quality and production efficiency of the product. At present, the ethanol production industry generally uses various sensors and monitoring equipment to detect the ethanol concentration in the production process in real time.

[0003] At present, traditional ethanol concentration monitoring methods mainly rely on concentration data fed back by sensors in real time. Although these methods can provide real-time concentration information during the production process, they often remain at the level of simple data collection and display, lacking in-depth analysis and mining of data. Production personnel can only manually judge whether the concentration is within the normal range by observing real-time data. This judgment method is not only inefficient, but also greatly affected by human factors and prone to misjudgment.

[0004] Most existing monitoring methods are unable to automatically analyze historical and real-time data to predict future trends in ethanol concentration. This means that production personnel cannot be informed of possible concentration anomalies in advance, and thus cannot take timely measures to adjust and optimize the production process. As a result, production personnel can only respond passively after problems occur, and cannot proactively prevent them from happening. Summary of the invention

[0005] The object of the present invention is to provide a method, device and system for online monitoring of ethanol production concentration to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: an online monitoring method for ethanol production concentration, the method comprising:

[0007] S1: Collect historical concentration data and related production parameters and environmental conditions during ethanol production; the concentration data includes ethanol concentration values ​​at different time points; the production parameters include raw material type, raw material ratio, reaction temperature, reaction pressure, catalyst type and reaction time; the environmental conditions include temperature, humidity and air pressure of the production environment;

[0008] S2: Preprocess the data collected in S1, including data cleaning, data alignment and missing value processing, to form a complete feature data set;

[0009] S3: Based on the feature data set, an ethanol concentration prediction model is constructed; the model combines production parameters and environmental conditions, and uses machine learning algorithms or deep learning algorithms to predict the concentration change trend in the future during the ethanol production process;

[0010] S4: Real-time acquisition of production status data and environmental condition data during the current ethanol production process;

[0011] S5: inputting the data obtained in S4 into the ethanol concentration prediction model, and receiving the predicted ethanol concentration value output by the model;

[0012] S6: Install an ethanol concentration sensor on the ethanol production line to detect the ethanol concentration in real time and obtain the actual ethanol concentration value output by the sensor; clean the original detection data to remove abnormal values ​​and noise interference;

[0013] S7: Plot the real-time ethanol concentration value obtained by S6 and the predicted ethanol concentration value output by the S5 model into time-varying curves, align the timestamps of the two curves, and put the two curves into the same coordinate system for comparison;

[0014] S8: According to a preset concentration deviation threshold, determine whether each point on the two curves has a deviation exceeding the threshold; if there is a deviation exceeding the threshold, mark the point as a deviation point, and record the deviation value and deviation time;

[0015] S9: Evaluate the concentration control status in the ethanol production process based on the number of deviation points found, the deviation value and the preset evaluation rules; if the evaluation result shows that there is an abnormality in the concentration control, the alarm mechanism is triggered, and the evaluation result and alarm information are sent to relevant personnel.

[0016] Preferably, in step S2, the collected ethanol concentration data, production parameters and environmental condition data are cleaned to remove duplicate data, erroneous data and abnormal data; the erroneous data include data errors caused by equipment failure, data transmission errors or human operation errors; the abnormal data include extreme data that exceeds a preset range or does not conform to production rules.

[0017] Preferably, in step S2, the data alignment is implemented by time-aligning the cleaned data, and aligning the data collected at different frequencies using an interpolation method or a resampling technique.

[0018] Preferably, a long short-term memory network LSTM algorithm is used to train the ethanol concentration prediction model, and the steps include:

[0019] Step 1: Divide the feature data set after S2 preprocessing into a training set, a validation set and a test set; the feature data set includes the ethanol concentration value of the time series, the numerical representation of the production parameters, and the numerical value of the environmental conditions; perform one-hot encoding processing on the categorical variables in the production parameters;

[0020] Step 2: Construct a long short-term memory network LSTM model, including an input layer, an LSTM layer, a fully connected layer, and an output layer; the input layer receives the preprocessed feature data; the LSTM layer consists of multiple LSTM units, each of which contains a forget gate, an input gate, and an output gate, and these three gating mechanisms are used to control the flow and memory of information; the fully connected layer is used to map the output of the LSTM layer to the prediction target; the output layer outputs the predicted ethanol concentration value;

[0021] Step 3: Use the training set to train the LSTM model, and use the back propagation algorithm and optimization algorithm to minimize the loss function; the loss function is used to measure the difference between the model's predicted value and the actual value; during the training process, the performance of the model is monitored through the validation set to prevent overfitting; the early stopping method is used to terminate the training, and the training is stopped when the loss on the validation set no longer decreases;

[0022] Step 4: Optimize the hyperparameters of the LSTM model using the grid search method, including the number of LSTM layers, the number of neurons in each layer, the learning rate, and the batch size;

[0023] Step 5: Evaluate the performance of the trained LSTM model on the test set and calculate the error index between the predicted value and the actual value;

[0024] Step 6: Deploy the trained LSTM model to the ethanol production line.

[0025] Preferably, two LSTM layers are designed, each layer contains 128 LSTM units; the first LSTM layer receives the output of the input layer, the second LSTM layer receives the output of the first LSTM layer, and a dropout layer is added between the two LSTM layers, and the dropout rate is set to 0.2.

[0026] Preferably, in step 5, the mean square error (MSE) is used as an evaluation indicator to calculate the average value of the square of the error between the model prediction value and the actual value; the calculation formula of MSE is as follows:

[0027]

[0028] Where n is the number of test set samples, y i is the actual ethanol concentration value of the i-th sample, is the ethanol concentration value of the i-th sample predicted by the model;

[0029] The prediction performance of the LSTM model is evaluated based on the calculated MSE value. The smaller the MSE value, the more accurate the model's prediction is and the smaller the deviation between the predicted value and the actual value is.

[0030] Preferably, in S8, the method for judging the ethanol concentration deviation is:

[0031] Set concentration deviation threshold: According to the actual demand and historical data of ethanol production process, set a reasonable concentration deviation threshold, which is used to judge whether the difference between the predicted concentration value and the actual concentration value is acceptable;

[0032] Deviation calculation: For each time point, the absolute deviation between the predicted concentration value and the actual concentration value is calculated; the calculation formula is: Deviation value = |predicted concentration value - actual concentration value|;

[0033] Deviation point judgment: Compare the calculated deviation value with the concentration deviation threshold. If the deviation value is greater than or equal to the concentration deviation threshold, mark the point as a deviation point and record the deviation value and deviation time.

[0034] Preferably, in S9, the ethanol concentration assessment and alarm mechanism design is as follows:

[0035] Count the number of deviation points found within a period of time and calculate the average deviation value or maximum deviation value of all deviation points.

[0036] According to the actual requirements of the ethanol production process, evaluation rules are set, including setting that when the number of deviation points exceeds a certain number or the average deviation value exceeds a certain threshold, it is considered that the concentration control is abnormal;

[0037] The concentration control status in the ethanol production process is judged based on the statistical number of deviation points and deviation values, as well as the set evaluation rules. When the evaluation results show that there is an abnormality in the concentration control, the alarm mechanism is triggered.

[0038] Preferably, an ethanol production concentration online monitoring system comprises:

[0039] A data acquisition module is used to execute step S1 to collect historical concentration data and related production parameters and environmental conditions during the ethanol production process, wherein the concentration data includes ethanol concentration values ​​at different time points, the production parameters include raw material type, raw material ratio, reaction temperature, reaction pressure, catalyst type and reaction time, and the environmental conditions include temperature, humidity and air pressure of the production environment;

[0040] The data preprocessing module is connected to the data acquisition module and is used to execute step S2 to preprocess the data collected by the data acquisition module, including data cleaning, data alignment and missing value processing, to form a complete feature data set;

[0041] A model building module is connected to the data preprocessing module and is used to execute step S3 to build an ethanol concentration prediction model based on the feature data set. The model combines production parameters and environmental conditions and uses a machine learning algorithm or a deep learning algorithm to predict the concentration change trend in the future during the ethanol production process.

[0042] A real-time data acquisition module is used to execute step S4 to acquire production status data and environmental condition data in the current ethanol production process in real time;

[0043] A prediction module, connected to the model building module and the real-time data acquisition module, is used to execute step S5, input the data acquired by the real-time data acquisition module into the ethanol concentration prediction model, and receive the predicted ethanol concentration value output by the model;

[0044] The sensor module is installed on the ethanol production line and is used to execute the first half of step S6, detect the ethanol concentration in real time, and obtain the actual ethanol concentration value output by the sensor;

[0045] A data cleaning module is connected to the sensor module and is used to execute the second half of step S6 to clean the original detection data output by the sensor module to remove abnormal values ​​and noise interference;

[0046] A comparison module is connected to the prediction module and the data cleaning module, and is used to execute steps S7 and S8, respectively plotting the real-time ethanol concentration value obtained by the data cleaning module and the predicted ethanol concentration value output by the prediction module into time-varying curves, aligning the timestamps of the two curves, and placing the two curves in the same coordinate system for comparison, and judging whether each point on the two curves has a deviation exceeding the threshold value according to a preset concentration deviation threshold value, and if there is a deviation exceeding the threshold value, marking the point as a deviation point, and recording the deviation value and deviation time;

[0047] The evaluation and alarm module is connected to the comparison module and is used to execute step S9. According to the number of deviation points found by the comparison module, the deviation value and the preset evaluation rules, the concentration control status in the ethanol production process is evaluated. If the evaluation result shows that there is an abnormality in the concentration control, the alarm mechanism is triggered and the evaluation result and alarm information are sent to relevant personnel.

[0048] Preferably, an online monitoring device for ethanol production concentration includes ethanol production equipment integrated with the above-mentioned online monitoring system for ethanol production concentration.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention constructs an ethanol concentration prediction model by using the long short-term memory network (LSTM) algorithm. The model can combine production parameters and environmental conditions to intelligently predict the trend of ethanol concentration changes in the future, so that production personnel can be informed of potential concentration anomalies in advance, and take timely measures to adjust and optimize during the production process, effectively preventing the occurrence of problems and improving production efficiency.

[0051] Enhance the initiative and accuracy of concentration control: By comparing and analyzing the real-time detected ethanol concentration value with the predicted value, the present invention can discover and mark the deviation point in real time, record the deviation value and deviation time, which not only improves the accuracy of concentration control, but also enables production personnel to actively respond to concentration changes and reduce product quality problems caused by concentration fluctuations.

[0052] Providing scientific evaluation and decision support: Based on the number of deviation points found, the deviation value and the preset evaluation rules, the present invention can scientifically evaluate the concentration control state in the ethanol production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A step diagram of an online monitoring method for ethanol production concentration according to the present invention;

[0054] Figure 2 This is a training flow chart for the ethanol concentration prediction model;

[0055] Figure 3 Schematic diagram of the design of ethanol concentration deviation judgment, evaluation and alarm mechanism. DETAILED DESCRIPTION

[0056] 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] See also Figure 1-3 The present invention provides an online monitoring method for ethanol production concentration, the method comprising:

[0058] S1: Collect historical concentration data and related production parameters and environmental conditions during the ethanol production process. Specifically, the concentration data includes ethanol concentration values ​​at different time points, the production parameters include raw material type, raw material ratio, reaction temperature, reaction pressure, catalyst type and reaction time, and the environmental conditions include temperature, humidity and air pressure of the production environment.

[0059] S2: Preprocess the data collected in S1. The preprocessing process includes removing duplicate data, erroneous data and abnormal data; the erroneous data includes data errors caused by equipment failure, data transmission errors or human operation errors; the abnormal data includes extreme data that exceeds the preset range or does not conform to production rules; data alignment to ensure the consistency of data from different sources in time; and missing value processing to reasonably fill or interpolate the missing data to form a complete feature data set.

[0060] Data alignment is achieved by time-aligning the cleaned data and aligning the data collected at different frequencies using interpolation methods or resampling techniques.

[0061] S3: Based on the preprocessed feature data set, an ethanol concentration prediction model is constructed. The model combines production parameters and environmental conditions, and uses deep learning algorithms (such as long short-term memory networks) to predict the concentration trend in the future during the ethanol production process. The model construction process includes steps such as data partitioning, feature selection, model training, and parameter optimization to ensure the accuracy and stability of the model.

[0062] S4: Real-time acquisition of production status data and environmental condition data during the current ethanol production process. These data are also collected in real time through sensors and monitoring equipment and input into the ethanol concentration prediction model.

[0063] S5: Input the real-time acquired data into the ethanol concentration prediction model, and receive the predicted ethanol concentration value output by the model. The predicted value represents the prediction result of the model for the ethanol concentration in the future.

[0064] S6: Install an ethanol concentration sensor on the ethanol production line to detect the ethanol concentration in real time and obtain the actual ethanol concentration value output by the sensor. Clean the original detection data to remove abnormal values ​​and noise interference to ensure the accuracy of the data.

[0065] S7: Plot the real-time ethanol concentration value and the model-predicted ethanol concentration value into time-varying curves, and align the timestamps. Put the two curves into the same coordinate system for comparison, so as to intuitively observe the difference between the actual concentration and the predicted concentration.

[0066] S8: According to the preset concentration deviation threshold, determine whether each point on the two curves has a deviation exceeding the threshold. If there is a deviation exceeding the threshold, mark the point as a deviation point, and record the deviation value and deviation time. This helps to detect abnormal conditions in concentration control in a timely manner.

[0067] S9: Evaluate the concentration control status in the ethanol production process based on the number of deviation points found, the deviation value and the preset evaluation rules. If the evaluation results show that the concentration control is abnormal, the alarm mechanism is triggered and the evaluation results and alarm information are sent to relevant personnel. Relevant personnel can take timely measures to adjust and optimize according to the alarm information, thereby ensuring the stability of the ethanol production process and product quality.

[0068] The present invention will be further described below in conjunction with Examples 1 to 3:

[0069] Embodiment 1:

[0070] See attached Figure 2 This example is used to describe a specific implementation method of using a long short-term memory network (LSTM) algorithm to train an ethanol concentration prediction model. The detailed steps are as follows:

[0071] Step 1: Divide the feature data set after S2 preprocessing into a training set (70%), a validation set (15%) and a test set (15%) according to the time series. The feature data set contains the ethanol concentration values ​​of the time series, the numerical representation of the production parameters (such as temperature, pressure, etc.), and the numerical values ​​of the environmental conditions (such as humidity, temperature, etc.).

[0072] For categorical variables in production parameters (such as raw material type, catalyst type, etc.), One-Hot Encoding is used to process them and convert them into numerical representation.

[0073] Step 2: Build a two-layer LSTM network model. The model includes an input layer, two LSTM layers, a fully connected layer, and an output layer.

[0074] The input layer receives the preprocessed feature data, and the data dimension is (time step, number of features).

[0075] The first LSTM layer contains 128 LSTM units, which receive the output of the input layer. The internal structure of the LSTM unit includes the forget gate, input gate, and output gate, which control the flow and memory of information through these three gating mechanisms. The forget gate determines the degree of retention of old information, the input gate determines the degree of addition of new information, and the output gate determines the output information based on the current cell state. The calculation method of the LSTM unit is:

[0076] Forget gate: f t =σ(W f ·[h t-1 ,x t ]+b f );

[0077] Input gate: i t =σ(W i ·[h t-1 ,x t ]+b i );

[0078] Output gate: o t =σ(W o ·[h t-1 ,x t ]+b o );

[0079] Among them, W and b represent the weight matrix and bias term respectively, and σ represents the sigmoid activation function.

[0080] A dropout layer is added after the first LSTM layer, and the dropout rate is set to 0.2 to prevent overfitting.

[0081] The second LSTM layer also contains 128 LSTM units and receives the output of the first LSTM layer.

[0082] The fully connected layer is used to map the output of the second LSTM layer to the predicted target, that is, the ethanol concentration value.

[0083] The output layer outputs the predicted ethanol concentration value using a linear activation function.

[0084] Step 3: Use the training set to train the LSTM model and minimize the loss function through the back propagation algorithm and optimization algorithm (such as Adam optimizer). The loss function uses the mean square error (MSE) to measure the difference between the model's predicted value and the actual value. The calculation formula of MSE is as follows:

[0085]

[0086] Where n is the number of test set samples, y i is the actual ethanol concentration value of the i-th sample, is the ethanol concentration value of the i-th sample predicted by the model.

[0087] The prediction performance of the LSTM model is evaluated based on the calculated MSE value. The smaller the MSE value, the more accurate the model's prediction is and the smaller the deviation between the predicted value and the actual value is.

[0088] During the training process, the validation set is used to monitor the performance of the model to prevent overfitting. The early stopping method is used to terminate the training, and the training is stopped when the loss on the validation set no longer decreases.

[0089] Step 4: Optimize the hyperparameters of the LSTM model through the grid search method, including the number of LSTM layers (fixed to 2 layers), the number of neurons in each layer (fixed to 128), the learning rate (such as 0.001, 0.0001, etc.) and the batch size (such as 32, 64, etc.).

[0090] During the grid search process, each hyperparameter combination is trained, and the performance is evaluated on the validation set to select the hyperparameter combination with the best performance.

[0091] Step 5: Evaluate the performance of the trained LSTM model on the test set and calculate the error metrics between the predicted values ​​and the actual values, such as accuracy and F1 score.

[0092] The prediction accuracy and generalization ability of the model are evaluated by error indicators.

[0093] Step 6: Deploy the trained LSTM model to the ethanol production line, receive production data in real time, and output the predicted ethanol concentration value.

[0094] According to the predicted ethanol concentration value, the production parameters are adjusted to optimize the ethanol production process.

[0095] Embodiment 2:

[0096] See attached Figure 3 In the specific implementation of the online monitoring method for ethanol production concentration, this embodiment is used to describe in detail the implementation method of the deviation judgment and evaluation mechanism:

[0097] ① The implementation methods of the deviation judgment mechanism include:

[0098] Set concentration deviation threshold: Based on historical data and ethanol production requirements, the concentration deviation threshold is set at 5% (i.e., a difference of more than 5% between the predicted concentration value and the actual concentration value is considered a deviation).

[0099] Deviation calculation: For each time point, the system calculates the absolute deviation between the predicted concentration value and the actual concentration value. For example, at a certain time point, the predicted concentration is 70% and the actual concentration is 65%, then the deviation value is |70%-65%|=5%.

[0100] Deviation point judgment: The system compares the calculated deviation value with the concentration deviation threshold. If the deviation value is greater than or equal to 5%, the system marks the point as a deviation point and records the specific deviation value and deviation time.

[0101] ② The implementation methods of the evaluation and alarm mechanism include:

[0102] Deviation point statistics: The system counts the number of deviation points found in a production cycle (such as 24 hours). For example, the system finds that there are 8 deviation points in a production cycle.

[0103] Deviation analysis: The system calculates the average deviation value of all deviation points. For example, the deviation values ​​of 8 deviation points are 5%, 6%, 7%, 5.5%, 4.8%, 5.2%, 6.1%, 5.9%, then the average deviation value is (5+6+...+5.9) / 8=5.6875%.

[0104] Evaluation rule setting: According to production requirements, the evaluation rule is set as follows: when the number of deviation points exceeds 5 or the average deviation value exceeds 5%, it is considered that there is an abnormality in the concentration control.

[0105] Evaluation result judgment: The system judges the concentration control status based on the number of statistical deviation points and the average deviation value, as well as the set evaluation rules. In this example, since the number of deviation points is 8 (greater than 5) and the average deviation value is 5.6875% (greater than 5%), the system judges that there is an abnormality in the concentration control.

[0106] Alarm mechanism triggering: When the system determines that there is an abnormality in concentration control, the alarm mechanism is triggered. The alarm mechanism includes: sending an alarm SMS to the production supervisor, emailing the technical department, and displaying a red alarm prompt on the monitoring interface.

[0107] Evaluation report generation: The system automatically generates an evaluation report, which includes evaluation results, alarm information, specific concentration data, deviation point information (including deviation time, deviation value, etc.). The evaluation report is saved in PDF format and can be downloaded or printed through the system interface.

[0108] Embodiment 3:

[0109] This embodiment is used to describe an ethanol production concentration online monitoring system and its implementation device, which includes the following main modules:

[0110] Data acquisition module: This module is responsible for collecting historical concentration data and related production parameters and environmental conditions during the ethanol production process. Concentration data includes ethanol concentration values ​​at different time points, production parameters include raw material type, raw material ratio, reaction temperature, reaction pressure, catalyst type and reaction time, and environmental conditions include temperature, humidity and air pressure of the production environment. These data provide the basis for subsequent data analysis and model building.

[0111] Data preprocessing module: connected to the data acquisition module, this module preprocesses the collected data, including data cleaning (removing errors and invalid data), data alignment (ensuring that the timestamps of each data point are consistent) and missing value processing (filling or estimating missing data), and finally forms a complete feature data set to provide high-quality data input for subsequent model construction.

[0112] Model building module: Based on the preprocessed feature data set, this module uses machine learning algorithms or deep learning algorithms to build an ethanol concentration prediction model. The model can combine production parameters and environmental conditions to predict the concentration trend in the future during the ethanol production process, providing a scientific basis for production decisions.

[0113] Real-time data acquisition module: This module acquires the production status data and environmental condition data of the current ethanol production process in real time, including but not limited to real-time parameters such as raw material flow, reaction temperature, pressure, and information such as ambient temperature and humidity, to provide data support for real-time prediction.

[0114] Prediction module: The data acquired by the real-time data acquisition module is input into the ethanol concentration prediction model, and the model outputs the predicted ethanol concentration value. These predicted values ​​reflect the expected changes in ethanol concentration in the future under the current production conditions.

[0115] Sensor module: Installed on the ethanol production line, it detects the ethanol concentration in real time and outputs the actual ethanol concentration value. These data are important bases for verifying the accuracy of the prediction model and monitoring the production status.

[0116] Data cleaning module: cleans the original detection data output by the sensor module, removes outliers and noise interference, and ensures the accuracy and reliability of the data.

[0117] Comparison module: The real-time ethanol concentration value obtained by the data cleaning module and the predicted ethanol concentration value output by the prediction module are plotted as time-varying curves, and the timestamps are aligned. The two curves are compared in the same coordinate system, and according to the preset concentration deviation threshold, it is determined whether each point has a deviation exceeding the threshold. If there is a deviation exceeding the threshold, it is marked as a deviation point, and the deviation value and deviation time are recorded.

[0118] Evaluation and alarm module: Based on the number of deviation points found by the comparison module, the deviation value and the preset evaluation rules, the concentration control status in the ethanol production process is evaluated. If the evaluation result shows that there is an abnormality in the concentration control, the alarm mechanism is triggered and the evaluation result and alarm information are sent to relevant personnel so that timely measures can be taken to adjust production conditions or troubleshoot.

[0119] In terms of implementation device, this embodiment provides an ethanol production concentration online monitoring device, which integrates the above-mentioned ethanol production concentration online monitoring system and is applied to ethanol production equipment. Through this device, real-time monitoring and prediction of ethanol production concentration can be achieved, improving the controllability and stability of the production process, thereby improving production efficiency and product quality.

[0120] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0121] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for online monitoring of ethanol production concentration, characterized in that: The method comprises: S1: Collect historical concentration data and related production parameters and environmental conditions during ethanol production; the concentration data includes ethanol concentration values ​​at different time points; the production parameters include raw material type, raw material ratio, reaction temperature, reaction pressure, catalyst type and reaction time; the environmental conditions include temperature, humidity and air pressure of the production environment; S2: Preprocess the data collected in S1, including data cleaning, data alignment and missing value processing, to form a complete feature data set; S3: Based on the feature data set, an ethanol concentration prediction model is constructed; the model combines production parameters and environmental conditions, and uses machine learning algorithms or deep learning algorithms to predict the concentration change trend in the future during the ethanol production process; S4: Real-time acquisition of production status data and environmental condition data during the current ethanol production process; S5: inputting the data obtained in S4 into the ethanol concentration prediction model, and receiving the predicted ethanol concentration value output by the model; S6: Install an ethanol concentration sensor on the ethanol production line to detect the ethanol concentration in real time and obtain the actual ethanol concentration value output by the sensor; clean the original detection data to remove abnormal values ​​and noise interference; S7: Plot the real-time ethanol concentration value obtained by S6 and the predicted ethanol concentration value output by the S5 model into time-varying curves, align the timestamps of the two curves, and put the two curves into the same coordinate system for comparison; S8: According to a preset concentration deviation threshold, determine whether each point on the two curves has a deviation exceeding the threshold; if there is a deviation exceeding the threshold, mark the point as a deviation point, and record the deviation value and deviation time; S9: Evaluate the concentration control status in the ethanol production process based on the number of deviation points found, the deviation value and the preset evaluation rules; if the evaluation result shows that there is an abnormality in the concentration control, the alarm mechanism is triggered, and the evaluation result and alarm information are sent to relevant personnel.

2. The method for online monitoring of ethanol production concentration according to claim 1, characterized in that: In step S2, the collected ethanol concentration data, production parameter and environmental condition data are cleaned to remove duplicate data, erroneous data and abnormal data; the erroneous data include data errors caused by equipment failure, data transmission errors or human operation errors; the abnormal data include extreme data that exceeds a preset range or does not conform to production rules.

3. The method for online monitoring of ethanol production concentration according to claim 2, characterized in that: In step S2, the data alignment is implemented by performing time alignment on the cleaned data, and using an interpolation method or a resampling technique to align data collected at different frequencies.

4. The method for online monitoring of ethanol production concentration according to claim 1, characterized in that: The ethanol concentration prediction model is trained using the long short-term memory network LSTM algorithm, and the steps include: Step 1: Divide the feature data set after S2 preprocessing into a training set, a validation set and a test set; the feature data set includes the ethanol concentration value of the time series, the numerical representation of the production parameters, and the numerical value of the environmental conditions; perform one-hot encoding processing on the categorical variables in the production parameters; Step 2: Construct a long short-term memory network LSTM model, including an input layer, an LSTM layer, a fully connected layer, and an output layer; the input layer receives the preprocessed feature data; the LSTM layer consists of multiple LSTM units, each of which contains a forget gate, an input gate, and an output gate, and these three gating mechanisms are used to control the flow and memory of information; the fully connected layer is used to map the output of the LSTM layer to the prediction target; the output layer outputs the predicted ethanol concentration value; Step 3: Use the training set to train the LSTM model, and use the back propagation algorithm and optimization algorithm to minimize the loss function; the loss function is used to measure the difference between the model's predicted value and the actual value; during the training process, the performance of the model is monitored through the validation set to prevent overfitting; the early stopping method is used to terminate the training, and the training is stopped when the loss on the validation set no longer decreases; Step 4: Optimize the hyperparameters of the LSTM model using the grid search method, including the number of LSTM layers, the number of neurons in each layer, the learning rate, and the batch size; Step 5: Evaluate the performance of the trained LSTM model on the test set and calculate the error index between the predicted value and the actual value; Step 6: Deploy the trained LSTM model to the ethanol production line.

5. The method for online monitoring of ethanol production concentration according to claim 4, characterized in that: Two LSTM layers are designed, each containing 128 LSTM units; the first LSTM layer receives the output of the input layer, and the second LSTM layer receives the output of the first LSTM layer. A dropout layer is added between the two LSTM layers, and the dropout rate is set to 0.

2.

6. The method for online monitoring of ethanol production concentration according to claim 4, characterized in that: In step 5, the mean square error (MSE) is used as an evaluation indicator to calculate the average value of the square of the error between the model prediction value and the actual value; the calculation formula of MSE is as follows: Where n is the number of test set samples, y i is the actual ethanol concentration value of the i-th sample, is the ethanol concentration value of the i-th sample predicted by the model; The prediction performance of the LSTM model is evaluated based on the calculated MSE value. The smaller the MSE value, the more accurate the model's prediction is and the smaller the deviation between the predicted value and the actual value is.

7. The method for online monitoring of ethanol production concentration according to claim 1, characterized in that: In S8, the method for judging the ethanol concentration deviation is: Set concentration deviation threshold: According to the actual demand and historical data of ethanol production process, set a reasonable concentration deviation threshold, which is used to judge whether the difference between the predicted concentration value and the actual concentration value is acceptable; Deviation calculation: For each time point, the absolute deviation between the predicted concentration value and the actual concentration value is calculated; the calculation formula is: Deviation value = |predicted concentration value - actual concentration value|; Deviation point judgment: Compare the calculated deviation value with the concentration deviation threshold. If the deviation value is greater than or equal to the concentration deviation threshold, mark the point as a deviation point and record the deviation value and deviation time.

8. The method for online monitoring of ethanol production concentration according to claim 1, characterized in that: In S9, the ethanol concentration assessment and alarm mechanism design is as follows: Count the number of deviation points found in a period of time, and calculate the average deviation value or maximum deviation value of all deviation points; According to the actual requirements of the ethanol production process, evaluation rules are set, including setting that when the number of deviation points exceeds a certain number or the average deviation value exceeds a certain threshold, it is considered that the concentration control is abnormal; The concentration control status in the ethanol production process is judged based on the statistical number of deviation points and deviation values, as well as the set evaluation rules. When the evaluation results show that there is an abnormality in the concentration control, the alarm mechanism is triggered.

9. An ethanol production concentration online monitoring system, characterized in that: The system comprises: A data acquisition module is used to execute step S1 to collect historical concentration data and related production parameters and environmental conditions during the ethanol production process, wherein the concentration data includes ethanol concentration values ​​at different time points, the production parameters include raw material type, raw material ratio, reaction temperature, reaction pressure, catalyst type and reaction time, and the environmental conditions include temperature, humidity and air pressure of the production environment; The data preprocessing module is connected to the data acquisition module and is used to execute step S2 to preprocess the data collected by the data acquisition module, including data cleaning, data alignment and missing value processing, to form a complete feature data set; A model building module is connected to the data preprocessing module and is used to execute step S3 to build an ethanol concentration prediction model based on the feature data set. The model combines production parameters and environmental conditions and uses a machine learning algorithm or a deep learning algorithm to predict the concentration change trend in the future during the ethanol production process. A real-time data acquisition module is used to execute step S4 to acquire production status data and environmental condition data in the current ethanol production process in real time; A prediction module, connected to the model building module and the real-time data acquisition module, is used to execute step S5, input the data acquired by the real-time data acquisition module into the ethanol concentration prediction model, and receive the predicted ethanol concentration value output by the model; The sensor module is installed on the ethanol production line and is used to execute the first half of step S6, detect the ethanol concentration in real time, and obtain the actual ethanol concentration value output by the sensor; A data cleaning module is connected to the sensor module and is used to execute the second half of step S6 to clean the original detection data output by the sensor module to remove abnormal values ​​and noise interference; A comparison module is connected to the prediction module and the data cleaning module, and is used to execute steps S7 and S8, respectively plotting the real-time ethanol concentration value obtained by the data cleaning module and the predicted ethanol concentration value output by the prediction module into time-varying curves, aligning the timestamps of the two curves, and placing the two curves in the same coordinate system for comparison, and judging whether each point on the two curves has a deviation exceeding the threshold value according to a preset concentration deviation threshold value, and if there is a deviation exceeding the threshold value, marking the point as a deviation point, and recording the deviation value and deviation time; The evaluation and alarm module is connected to the comparison module and is used to execute step S9. According to the number of deviation points found by the comparison module, the deviation value and the preset evaluation rules, the concentration control status in the ethanol production process is evaluated. If the evaluation result shows that there is an abnormality in the concentration control, the alarm mechanism is triggered and the evaluation result and alarm information are sent to relevant personnel.

10. An online monitoring device for ethanol production concentration, characterized in that: An ethanol production device comprising an ethanol production concentration online monitoring system as described in claim 9 is integrated.

Images