System and method for optimally controlling heat exchanger through neural network

Abstract

The invention relates to the technical field of heat exchanger control, and particularly discloses a system and method for optimizing and controlling a heat exchanger through a neural network, the system comprises a data acquisition and processing system and a control decision system, a parameter acquisition module is used for collecting operation parameters of the heat exchanger, and a data preprocessing module is used for preprocessing the operation parameters; when the predicted flow is in a preset reasonable flow interval, the flow valve opening degree is adjusted according to a predicted value, if the predicted flow exceeds the reasonable flow interval, the size of an out-of-range is further judged, if the out-of-range is small, deviation information is recorded, and if the out-of-range is large, an alarm is triggered, and fault maintenance information is prompted. The temperature of a material outlet can be stable, great advantages are achieved for heat-sensitive materials, the inlet and outlet states of the heat exchanger can be monitored and predicted in real time, the purpose of real-time monitoring and alarming can be achieved, the safety in the industrial process can be improved, the inlet cooling water flow can be regulated and controlled in real time, and cooling water consumption can reach the optimal value.

Description

Technical Field

[0001] The present invention relates to the technical field of heat exchanger control, and in particular to a system and method for optimizing and controlling a heat exchanger using a neural network. Background Art

[0002] As an important equipment in industrial production, the main function of the heat exchanger is to achieve heat transfer between different media, so as to meet the demand for temperature control in the process. The main function of the heat exchanger is to achieve heat transfer, and it plays a vital role in the industrial process. By controlling the temperature, improving product quality and output, it is conducive to promoting the effective use of energy and environmental protection and emission reduction.

[0003] The heat exchangers designed in the current industry will also have a certain amount of heat exchange margin, but the materials that need to be cooled in industrial production are unstable, and the cooling medium flow rate is generally set to a rated value. The initial temperature of the cooling water will also have a certain deviation in a year. This will cause the temperature of the heat exchanger outlet to fluctuate. If the heat exchange material is sensitive to temperature, the heat exchanger designed in general cannot meet the demand, and if it is adjusted manually, it will require more labor costs.

[0004] Therefore, there is an urgent need for a system and method for optimizing the control of a heat exchanger using a neural network to solve the above problems. Summary of the invention

[0005] The object of the present invention is to provide a system for optimizing and controlling a heat exchanger using a neural network, comprising: a data acquisition and processing system and a control decision system, wherein the data acquisition and processing system comprises a parameter acquisition module, a data preprocessing module, a feature extraction module and a data transmission module, and the control decision system comprises a model training module, a flow prediction module and a control execution module;

[0006] The parameter acquisition module is used to collect the heat exchanger operating parameters, which include the inlet and outlet temperature, pressure, and flow information of the hot material and the inlet and outlet temperature and pressure information of the cooling water;

[0007] The data preprocessing module is connected to the parameter acquisition module, and the data preprocessing module is used to preprocess the acquired heat exchanger operating parameters to obtain preprocessed data;

[0008] The feature extraction module is connected to the data preprocessing module, and the feature extraction module is used to extract key feature information from the preprocessed data;

[0009] The data transmission module is connected to the feature extraction module, and the data transmission module is used to transmit the feature vector to the computing device;

[0010] The model training module is connected to the data transmission module, and the model training module is used to receive the feature vector and use it to train the neural network model, and obtain an optimized neural network prediction model;

[0011] The flow prediction module is connected to the model training module. During the actual operation of the heat exchanger, the flow prediction module inputs the feature vector collected and processed in real time into the neural network prediction model to predict the required cooling water flow data;

[0012] The control execution module is connected to the flow prediction module, and the control execution module is used to adjust and control the cooling water inlet flow of the heat exchanger according to the predicted cooling water flow;

[0013] If the predicted flow rate is within the preset reasonable flow rate range, the flow valve opening is adjusted according to the predicted value;

[0014] If it exceeds the reasonable flow range, the size of the excess range is further determined. If the excess range is small, fine-tuning is performed and the deviation information is recorded. If the excess range is large, an alarm is triggered and fault repair information is prompted.

[0015] Furthermore, the parameter acquisition module is used to obtain the inlet and outlet temperature, pressure, flow rate of hot materials and the inlet and outlet temperature and pressure parameters of cooling water of the heat exchanger in real time through temperature sensors, pressure sensors and flow sensors.

[0016] Furthermore, the step of obtaining preprocessed data includes data cleaning and normalization processing, removing abnormal data and converting the data to a unified dimension. The data preprocessing module uses a data filtering algorithm to remove noise interference in the collected data, and uses a standardized method to normalize parameter data of different magnitudes to a specific range.

[0017] Furthermore, the step of extracting key feature information includes combining the extracted key features into a feature vector, the key features include the temperature change rate of the hot material, the cooling water temperature change trend, and the hot material flow fluctuation characteristics. The feature extraction module extracts the change rate of the hot material temperature within a certain period of time based on the time series analysis method, determines the rising or falling trend of the cooling water temperature through the trend analysis algorithm, calculates the standard deviation and skewness characteristics of the hot material flow using statistical methods, and combines the standard deviation and skewness characteristics into a feature vector of the heat exchanger operating status.

[0018] Furthermore, the step of obtaining the optimized neural network prediction model includes training the neural network multiple times through historical data and simulation data, so that the neural network learns the complex mapping relationship between the heat exchanger operating parameters and the cooling water flow, so as to obtain the optimized neural network prediction model. The model training module uses a long short-term memory network to construct a neural network architecture, and uses a stochastic gradient descent optimization algorithm to train the neural network to obtain adjustment network weights and bias data.

[0019] Furthermore, assuming that the input feature vector is Corresponding to the hot material inlet temperature , Hot material inlet flow , Cooling water inlet temperature , hot material outlet temperature , then the formula for predicting water flow y is:

[0020] ;

[0021] in , , , is the weight coefficient and b is the bias term.

[0022] Further, assuming that the reasonable flow range is ;

[0023] when ≤y≤ When the flow valve opening is adjusted according to the predicted value;

[0024] when or When the value exceeds the specified range, it is considered that the value exceeds the specified range by a small amount, and fine-tuning is performed and the deviation information is recorded;

[0025] when or When the value exceeds the specified range, it is considered to be too large, triggering an alarm and prompting fault repair information;

[0026] The value range of α is 0<α<1, and the value range of β is β>1.

[0027] The present invention also provides a method for optimizing and controlling a heat exchanger using a neural network, comprising:

[0028] Obtain operating parameter information of the heat exchanger;

[0029] Preprocessing the collected operating parameter information of the heat exchanger to obtain preprocessing data;

[0030] Extract key feature information from preprocessed data;

[0031] transmitting the feature vector to a computing device;

[0032] Receive the feature vector and use it to train the neural network model, and obtain an optimized neural network prediction model;

[0033] Input the real-time collected and processed feature vectors into the neural network prediction model to predict the required cooling water flow data;

[0034] Adjust and control the cooling water inlet flow of the heat exchanger according to the predicted cooling water flow;

[0035] If the predicted flow rate is within the preset reasonable flow rate range, the flow valve opening is adjusted according to the predicted value;

[0036] If it exceeds the reasonable flow range, the size of the excess range is further determined. If the excess range is small, fine-tuning is performed and the deviation information is recorded. If the excess range is large, an alarm is triggered and fault repair information is prompted.

[0037] The present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0038] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0039] The beneficial effects of this application are:

[0040] First, the present invention establishes a data acquisition and processing system and a control decision system, and uses a neural network model to accurately predict and control the cooling water flow rate. The parameter acquisition module collects comprehensive heat exchanger operating parameters, providing a rich data basis for subsequent accurate control, thereby solving the problem of heat exchanger outlet temperature fluctuations caused by factors such as unstable cooling materials and deviations in cooling water temperature in industrial production.

[0041] Secondly, the present invention also abandons the traditional manual adjustment method, reducing labor costs. At the same time, through precise flow control and effective monitoring of the operating status of the heat exchanger, it avoids over-design of the heat exchange area of ​​the heat exchanger and unreasonable cooling water consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a schematic diagram of a method flow proposed in an embodiment of the present application.

[0043] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0044] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0045] Embodiment 1

[0046] The present application provides a system for optimizing and controlling a heat exchanger using a neural network, including: a data acquisition and processing system and a control decision system, wherein the data acquisition and processing system includes a parameter acquisition module, a data preprocessing module, a feature extraction module, and a data transmission module, and the control decision system includes a model training module, a flow prediction module, and a control execution module;

[0047] The parameter acquisition module is used to collect the heat exchanger operating parameters, which include the inlet and outlet temperature, pressure, and flow information of the hot material and the inlet and outlet temperature and pressure information of the cooling water;

[0048] The data preprocessing module is connected to the parameter acquisition module, and the data preprocessing module is used to preprocess the acquired heat exchanger operating parameters to obtain preprocessed data;

[0049] The feature extraction module is connected to the data preprocessing module, and the feature extraction module is used to extract key feature information from the preprocessed data;

[0050] The data transmission module is connected to the feature extraction module, and the data transmission module is used to transmit the feature vector to the computing device;

[0051] The model training module is connected to the data transmission module, and the model training module is used to receive the feature vector and use it to train the neural network model, and obtain an optimized neural network prediction model;

[0052] The flow prediction module is connected to the model training module. During the actual operation of the heat exchanger, the flow prediction module inputs the feature vector collected and processed in real time into the neural network prediction model to predict the required cooling water flow data;

[0053] The control execution module is connected to the flow prediction module, and the control execution module is used to adjust and control the cooling water inlet flow of the heat exchanger according to the predicted cooling water flow;

[0054] If the predicted flow rate is within the preset reasonable flow rate range, the flow valve opening is adjusted according to the predicted value;

[0055] If it exceeds the reasonable flow range, the size of the excess range is further determined. If the excess range is small, fine-tuning is performed and the deviation information is recorded. If the excess range is large, an alarm is triggered and fault repair information is prompted.

[0056] In the present invention, by establishing a data acquisition and processing system and a control decision system, a neural network model is used to accurately predict and control the cooling water flow rate. The parameter acquisition module collects comprehensive heat exchanger operating parameters, providing a rich data basis for subsequent accurate control, thereby being able to solve the problem of heat exchanger outlet temperature fluctuations due to factors such as unstable cooling materials and cooling water temperature deviations in industrial production. The cooling water flow rate can be adjusted in real time according to changes in operating parameters to keep the hot material outlet temperature stable, meeting the process requirements for temperature-sensitive materials and overcoming the defect that traditional heat exchangers cannot effectively stabilize the temperature when facing complex working conditions.

[0057] The present invention also abandons the traditional manual adjustment method and reduces labor costs. At the same time, through precise flow control and effective monitoring of the operating status of the heat exchanger, over-design of the heat exchange area of ​​the heat exchanger and unreasonable cooling water consumption are avoided. In the past, manual adjustment of the temperature of the heat exchanger required a lot of manpower and was costly. In addition, the traditional heat exchanger is designed to reserve a large heat exchange margin to deal with extreme situations, resulting in equipment costs and energy waste. The present invention realizes automatic intelligent control of the heat exchanger, optimizes the operating efficiency of the heat exchanger, and reduces equipment investment and operating costs. Compared with the traditional heat exchanger, which lacks an effective self-monitoring and fault warning mechanism during operation, when problems such as blockage and dirt occur, it is difficult to detect and deal with them in time, which may lead to production interruption or product quality degradation. The present invention can detect potential problems in advance, perform maintenance in time, and improve the reliability of the heat exchanger and the continuity of the production process. It also has real-time monitoring and fault warning functions for the operating status of the heat exchanger. The control execution module can detect abnormalities in time according to the flow prediction results and take corresponding measures to ensure the safe and stable operation of the equipment.

[0058] The present invention can stabilize the outlet temperature of the material, has great advantages for heat-sensitive materials, can monitor and predict the inlet and outlet status of the heat exchanger in real time, achieve the purpose of real-time detection and alarm, improve the safety of the industrial process, and can also regulate the inlet cooling water flow in real time to achieve the optimal cooling water consumption.

[0059] Specifically, the parameter acquisition module is used to obtain the inlet and outlet temperature, pressure, flow rate of hot materials and the inlet and outlet temperature and pressure parameters of cooling water of the heat exchanger in real time through temperature sensors, pressure sensors and flow sensors. To ensure the accuracy and timeliness of the collected data and provide a reliable basis for subsequent analysis and control, the inlet and outlet temperature, pressure, flow rate of hot materials and the inlet and outlet temperature and pressure parameters of cooling water of the heat exchanger can be obtained in real time through temperature sensors, pressure sensors and flow sensors. This real-time monitoring ensures the accuracy and timeliness of the collected data and provides a reliable basis for subsequent analysis and control, and accurate and timely data is crucial for subsequent analysis and control. With these data, the system can better understand the operating status of the heat exchanger and then perform effective optimization control, such as accurately regulating the cooling water flow through a neural network model to ensure efficient operation of the heat exchanger.

[0060] Specifically, the step of obtaining preprocessed data includes data cleaning and normalization, removing abnormal data and converting data to a unified dimension. The data preprocessing module uses a data filtering algorithm to remove noise interference in the collected data, and uses a standardization method to normalize parameter data of different magnitudes to a specific interval. For example, temperature data is normalized to the interval [0, 1], and pressure data is normalized to the interval [-1, 1], etc., to improve data quality and model training effect.

[0061] The present invention can remove abnormal data and noise interference in the collected data through data cleaning and normalization. This helps to ensure the accuracy and reliability of the data, and provides a high-quality data basis for subsequent analysis and control. The use of standardized methods to normalize parameter data of different magnitudes to a specific interval helps to eliminate the impact caused by magnitude differences between different parameters, so that the data is processed and analyzed on the same scale, improving the consistency and comparability of data processing. The pre-processed data can significantly improve the effect of model training because noise and outliers are removed and dimensional unification is achieved. For the neural network model trained based on these data, the laws and features in the data can be learned more accurately, thereby improving the prediction accuracy and stability of the model, and then improving the performance of the entire control system.

[0062] Specifically, the step of extracting key feature information includes combining the extracted key features into a feature vector, the key features include the temperature change rate of the hot material, the cooling water temperature change trend, and the hot material flow fluctuation characteristics. The feature extraction module extracts the change rate of the hot material temperature within a certain period of time based on the time series analysis method, determines the rising or falling trend of the cooling water temperature through the trend analysis algorithm, calculates the standard deviation and skewness characteristics of the hot material flow using a statistical method, and combines the standard deviation and skewness characteristics into a feature vector of the heat exchanger operating state.

[0063] The step of obtaining the optimized neural network prediction model includes training the neural network multiple times through historical data and simulation data, so that the neural network learns the complex mapping relationship between the operating parameters of the heat exchanger and the cooling water flow, so as to obtain the optimized neural network prediction model. The model training module adopts a long short-term memory network to construct a neural network architecture, and uses a stochastic gradient descent optimization algorithm to train the neural network to obtain and adjust network weights and bias data to minimize the error between the cooling water flow predicted by the model and the actual demand, thereby improving the accuracy and generalization ability of the model. The control execution module is connected to equipment such as flow control valves or pumps, and accurately controls the valve opening or the pump speed according to the predicted cooling water flow value to achieve precise regulation of the cooling water flow; when an alarm is triggered, the operator is prompted by an audible and visual alarm device, and fault-related information such as abnormal flow value and occurrence time are recorded in a log file to facilitate subsequent fault troubleshooting and analysis.

[0064] The present invention can accurately capture key information during the operation of the heat exchanger by extracting key features such as the rate of change of the hot material temperature, the trend of the cooling water temperature change, and the fluctuation characteristics of the hot material flow rate. These features can better reflect the operating status of the heat exchanger and provide a valuable data basis for subsequent model training and control. For example, in actual operation, the rate of change of the hot material temperature can reflect the heat transfer efficiency of the hot material in the heat exchanger, and the trend of the cooling water temperature change can help determine whether the cooling effect of the cooling water is normal.

[0065] Using historical data and simulation data to train the neural network multiple times enables the neural network to learn the complex mapping relationship between the heat exchanger operating parameters and the cooling water flow, which helps to improve the prediction accuracy of the model and enable it to more accurately predict the cooling water flow. For example, in industrial production, the operating parameters and cooling water flow requirements of the heat exchanger under different working conditions are different. The neural network model trained with a large amount of data can adapt to these changes and provide cooling water flow prediction values ​​that are more in line with actual needs.

[0066] The present invention uses a long short-term memory network (LSTM) to construct a neural network architecture, which is suitable for processing time series data and can effectively capture long-term dependencies in the data. At the same time, the neural network is trained using a stochastic gradient descent optimization algorithm, which can quickly adjust the network weights and bias data to minimize the error between the cooling water flow predicted by the model and the actual demand, thereby improving the accuracy and generalization ability of the model. For example, when processing data with time series characteristics such as heat exchanger operation data, the LSTM network can better handle the changes in parameters such as hot materials and cooling water temperature and flow over time, thereby providing more accurate prediction results. The stochastic gradient descent algorithm can efficiently optimize the model parameters during the training process, allowing the model to converge to the optimal solution more quickly.

[0067] Specifically, assume that the input feature vector is Corresponding to the hot material inlet temperature , Hot material inlet flow , Cooling water inlet temperature , hot material outlet temperature , then the formula for predicting water flow y is:

[0068] ;

[0069] in , , , is the weight coefficient and b is the bias term.

[0070] During the training process, based on a large amount of experimental or historical operation data, these weight coefficients and bias terms are continuously adjusted by minimizing the error (such as mean square error) between the predicted value y and the actual water flow value to improve the accuracy of the prediction.

[0071] Further, assuming that the reasonable flow range is ; Usually a relatively small ratio value α (0<α<1) and a relatively large ratio value β (β>1) can be set to define the degree of exceeding the range.

[0072] when ≤y≤ When the flow valve opening is adjusted according to the predicted value;

[0073] when or When the value exceeds the specified range, it is considered that the value exceeds the specified range by a small amount, and fine-tuning is performed and the deviation information is recorded;

[0074] when or When the value exceeds the specified range, it is considered to be too large, triggering an alarm and prompting fault repair information.

[0075] At this time, the fault indication variable F=1, and when it is within the normal range or exceeds the range by a small amount, F=0.

[0076] Here, α, β, and It needs to be determined according to the specific heat exchanger system, operating requirements and empirical data. For example, α can be 0.16, β can be 1.5, and and It should be obtained through experiments or theoretical calculations based on factors such as the design specifications of the heat exchanger, the properties of the hot material and cooling water, and the process requirements.

[0077] The present invention can accurately predict the cooling water flow required by the heat exchanger by establishing a water flow prediction formula based on input feature vectors (including hot material inlet temperature, hot material inlet flow, cooling water inlet temperature, and hot material outlet temperature). This multi-parameter-based prediction formula takes into account multiple key factors that affect the operation of the heat exchanger. Compared with a single-factor prediction method, it can more comprehensively reflect the actual operating status of the heat exchanger, thereby improving the accuracy of flow prediction.

[0078] During the training process, based on a large amount of experimental or historical operating data, the weight coefficient and bias term are adjusted by minimizing the error (such as mean square error) between the predicted value and the actual water flow value, so that the model can be continuously optimized and the prediction accuracy can be improved. This adaptive training method allows the model to continuously learn and adjust according to actual data, so that it can adapt to the operation of the heat exchanger under different working conditions, and enhance the generalization ability of the model.

[0079] The present invention sets a reasonable flow range and defines the degree of exceeding the range by a relatively small ratio α and a relatively large ratio β, and then adopts different control strategies (such as adjusting the flow valve opening according to the predicted value, fine-tuning and recording deviation information, triggering an alarm and prompting fault repair information). It can achieve refined control of the cooling water flow of the heat exchanger. This flexible control strategy can timely detect and handle abnormal situations while ensuring the normal operation of the heat exchanger, avoid problems such as reduced heat exchanger efficiency or equipment damage caused by excessive or insufficient flow, and improve the reliability and stability of the system.

[0080] like Figure 1 As shown, the present invention also provides a method for optimizing and controlling a heat exchanger using a neural network, comprising:

[0081] S1, obtaining operating parameter information of the heat exchanger;

[0082] S2, preprocessing the collected heat exchanger operating parameter information to obtain preprocessed data;

[0083] S3, extract key feature information from the preprocessed data;

[0084] S4, transmitting the feature vector to a computing device;

[0085] S5, receiving the feature vector and using it to train the neural network model, and obtaining an optimized neural network prediction model;

[0086] S6, inputting the feature vector collected and processed in real time into the neural network prediction model to predict the required cooling water flow data;

[0087] S7, regulating and controlling the cooling water inlet flow of the heat exchanger according to the predicted cooling water flow;

[0088] S8, if the predicted flow is within the preset reasonable flow range, the flow valve opening is adjusted according to the predicted value;

[0089] S9, if it exceeds the reasonable flow range, further determine the size of the excess range. If the excess range is small, fine-tune and record the deviation information. If the excess range is large, trigger an alarm and prompt fault repair information.

[0090] The present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0091] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0092] Embodiment 2

[0093] The difference between this embodiment and the first embodiment is that the present invention also includes an adaptive weight adjustment mechanism;

[0094] In the process of model training, in addition to training based on a large amount of experimental or historical operating data, an adaptive weight adjustment mechanism is added. That is, during the actual operation of the heat exchanger, the weight coefficient and bias term in the neural network model are adjusted in real time according to the deviation between the cooling water flow prediction result and the actual flow value each time. For example, a dynamic weight adjustment algorithm based on real-time feedback is adopted. The prediction error is calculated immediately after each prediction, and the weight is adjusted according to the error size and direction in a certain proportion. If the predicted flow is greater than the actual flow and the error exceeds a certain threshold, the weight coefficient related to the hot material inlet flow is reduced accordingly, and vice versa. This real-time adjustment mechanism can adapt to sudden changes in the operating conditions of the heat exchanger more quickly, such as sudden changes in the properties of the hot material or fluctuations in the cooling water supply, and further improve the accuracy and timeliness of the model prediction.

[0095] An online learning module is established to work together with the model training module and the flow prediction module. The online learning module can continuously collect and store the latest operating data of the heat exchanger and regularly add new data to the training set of the neural network model.

[0096] A data update cycle is set according to actual production needs. For example, after each hour or each batch of production, the newly collected data is mixed with the historical data to retrain the neural network model. During the retraining process, an incremental learning algorithm is used to avoid the huge amount of calculation caused by retraining the entire data set. Through the online learning mechanism, the model can continuously learn and adapt to the slow changes in the long-term operation of the heat exchanger, such as performance changes caused by equipment aging, pipeline scaling and other factors, to ensure that the model can maintain a high prediction accuracy throughout the life cycle of the heat exchanger.

[0097] Based on adaptive weight adjustment and online learning mechanism, the flow control strategy is optimized. When the cooling water flow predicted by the model exceeds the reasonable flow range, before fine-tuning or alarming, it is first determined whether it is caused by new working conditions that the model has not yet adapted to.

[0098] If the flow deviation is caused by a new operating condition, and the deviation is within a certain range, the adaptive weight adjustment mechanism is preferentially used to quickly correct the model, and then the flow is re-predicted and adjusted based on the corrected model to reduce unnecessary fine-tuning operations and false alarms. Only when the flow deviation is still large after multiple adaptive adjustments is it determined to be an abnormal situation that exceeds the range by a large amount, triggering an alarm and prompting fault repair information. By introducing adaptive weight adjustment and online learning mechanisms, this embodiment can further improve the accuracy of heat exchanger cooling water flow prediction and control, enhance the system's adaptability to various operating condition changes, and ensure the long-term stable and efficient operation of the heat exchanger.

[0099] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, value library or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0100] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0101] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent results or equivalent process changes made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A system for optimizing and controlling a heat exchanger using a neural network, characterized in that: include: Data acquisition and processing system and control decision system, wherein the data acquisition and processing system includes parameter acquisition module, data preprocessing module, feature extraction module and data transmission module, and the control decision system includes model training module, flow prediction module and control execution module; The parameter acquisition module is used to collect the heat exchanger operating parameters, which include the inlet and outlet temperature, pressure, and flow information of the hot material and the inlet and outlet temperature and pressure information of the cooling water; The data preprocessing module is connected to the parameter acquisition module, and the data preprocessing module is used to preprocess the acquired heat exchanger operating parameters to obtain preprocessed data; The feature extraction module is connected to the data preprocessing module, and the feature extraction module is used to extract key feature information from the preprocessed data; The data transmission module is connected to the feature extraction module, and the data transmission module is used to transmit the feature vector to the computing device; The model training module is connected to the data transmission module, and the model training module is used to receive the feature vector and use it to train the neural network model, and obtain an optimized neural network prediction model; The flow prediction module is connected to the model training module. During the actual operation of the heat exchanger, the flow prediction module inputs the feature vector collected and processed in real time into the neural network prediction model to predict the required cooling water flow data; The control execution module is connected to the flow prediction module, and the control execution module is used to adjust and control the cooling water inlet flow of the heat exchanger according to the predicted cooling water flow; If the predicted flow rate is within the preset reasonable flow rate range, the flow valve opening is adjusted according to the predicted value; If it exceeds the reasonable flow range, the size of the excess range is further determined. If the excess range is small, fine-tuning is performed and the deviation information is recorded. If the excess range is large, an alarm is triggered and fault repair information is prompted.

2. According to claim 1, a neural network optimization control system for heat exchangers is characterized in that: The parameter acquisition module is used to obtain the inlet and outlet temperature, pressure, flow rate of hot materials and the inlet and outlet temperature and pressure parameters of cooling water of the heat exchanger in real time through temperature sensors, pressure sensors and flow sensors.

3. The system for optimizing and controlling a heat exchanger using a neural network according to claim 1, characterized in that: The step of obtaining the preprocessed data includes data cleaning and normalization processing, removing abnormal data and converting the data to a unified dimension. The data preprocessing module uses a data filtering algorithm to remove noise interference in the collected data and uses a standardized method to normalize parameter data of different magnitudes to a specific range.

4. The system for optimizing and controlling a heat exchanger using a neural network according to claim 1, characterized in that: The step of extracting key feature information includes combining the extracted key features into a feature vector, wherein the key features include the temperature change rate of the hot material, the temperature change trend of the cooling water, and the fluctuation characteristics of the hot material flow rate. The feature extraction module extracts the temperature change rate of the hot material within a certain period of time based on a time series analysis method, determines the rising or falling trend of the cooling water temperature through a trend analysis algorithm, calculates the standard deviation and skewness characteristics of the hot material flow rate using a statistical method, and combines the standard deviation and skewness characteristics into a feature vector of the operating status of the heat exchanger.

5. The system for optimizing and controlling a heat exchanger using a neural network according to claim 4, characterized in that: The step of obtaining the optimized neural network prediction model includes training the neural network multiple times through historical data and simulation data, so that the neural network learns the complex mapping relationship between the heat exchanger operating parameters and the cooling water flow, so as to obtain the optimized neural network prediction model. The model training module uses a long short-term memory network to construct a neural network architecture, and uses a stochastic gradient descent optimization algorithm to train the neural network to obtain adjustment network weights and bias data.

6. The system for optimizing and controlling a heat exchanger using a neural network according to claim 5, characterized in that: Assume that the input feature vector is Corresponding to the hot material inlet temperature , Hot material inlet flow , Cooling water inlet temperature , hot material outlet temperature , then the formula for predicting water flow y is: ； in , , , is the weight coefficient and b is the bias term.

7. The system for optimizing and controlling a heat exchanger using a neural network according to claim 6, characterized in that: Assume that the reasonable flow range is ; when ≤y≤ When the flow valve opening is adjusted according to the predicted value; when or When the value exceeds the specified range, it is considered that the value exceeds the specified range by a small amount, and fine-tuning is performed and the deviation information is recorded; when or When the value exceeds the specified range, it is considered to be too large, triggering an alarm and prompting fault repair information; The value range of α is 0<α<1, and the value range of β is β>1.

8. A method for optimizing and controlling a heat exchanger using a neural network, characterized in that: include: Obtain operating parameter information of the heat exchanger; Preprocessing the collected operating parameter information of the heat exchanger to obtain preprocessing data; Extract key feature information from preprocessed data; transmitting the feature vector to a computing device; Receive the feature vector and use it to train the neural network model, and obtain an optimized neural network prediction model; Input the real-time collected and processed feature vectors into the neural network prediction model to predict the required cooling water flow data; Adjust and control the cooling water inlet flow of the heat exchanger according to the predicted cooling water flow; If the predicted flow rate is within the preset reasonable flow rate range, the flow valve opening is adjusted according to the predicted value; If it exceeds the reasonable flow range, the size of the excess range is further determined. If the excess range is small, fine-tuning is performed and the deviation information is recorded. If the excess range is large, an alarm is triggered and fault repair information is prompted.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for optimizing control of a heat exchanger by a neural network as recited in claim 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for optimizing control of a heat exchanger by a neural network as recited in claim 8 are implemented.

Images