A Method and System for Compensating the Data Accuracy of a Pressure Sensor for Industrial Control

Abstract

The present invention provides a method and system for compensating the data accuracy of a pressure sensor for industrial control. First, an original pressure signal sequence output by a target pressure sensor within a preset acquisition period is obtained. This original pressure signal sequence contains multiple pressure signals arranged according to timestamps. At the same time, a set of real-time environmental parameters during the acquisition period is detected, covering temperature, humidity, and mechanical vibration parameters. Then, the set of real-time environmental parameters is input into a pre-trained compensation parameter generation model to generate a corresponding set of dynamic compensation parameters. Subsequently, each signal in the original pressure signal sequence is adjusted step by step according to the set of dynamic compensation parameters to obtain a compensated pressure signal sequence. Finally, the compensated pressure signal sequence is output to the signal processing module of the industrial control system to drive the actuator, so that the data accuracy of the pressure sensor can be dynamically and accurately compensated by integrating multiple environmental factors, improving the accuracy and stability of industrial control.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular, to a method and system for compensating the data accuracy of a pressure sensor for industrial control. Background Art

[0002] In the process of industrial control, pressure sensors play a crucial role, and the accuracy of their measurement data directly affects the stability of industrial production, product quality, and the reliability of the entire system. With the continuous improvement of industrial automation, the requirements for the data accuracy of pressure sensors are becoming increasingly strict.

[0003] Traditional methods for processing pressure sensor data are relatively simple and usually only rely on the fixed calibration parameters set at the factory of the sensor itself to correct the measurement data. This method does not fully consider the influence of complex and variable factors in the actual industrial environment on the measurement accuracy of the pressure sensor. In the actual industrial scenario, the environment where the pressure sensor is located is often very complex, and factors such as temperature fluctuations, humidity changes, and mechanical vibrations may cause deviations in the pressure signals output by the sensor, thereby causing errors in industrial control based on these inaccurate signals. In severe cases, it may even lead to production accidents or a decrease in product quality. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide a method for compensating the data accuracy of a pressure sensor for industrial control, the method comprising:

[0005] Obtain an original pressure signal sequence output by a target pressure sensor within a preset acquisition period, where the original pressure signal sequence contains a plurality of pressure signals arranged according to timestamps;

[0006] Detect a set of real-time environmental parameters of the target pressure sensor within the acquisition period, where the set of real-time environmental parameters includes temperature parameters, humidity parameters, and mechanical vibration parameters;

[0007] Input the set of real-time environmental parameters into a pre-trained compensation parameter generation model to generate a set of dynamic compensation parameters corresponding to the original pressure signal sequence;

[0008] According to the set of dynamic compensation parameters, perform step-by-step adjustment on each pressure signal in the original pressure signal sequence to obtain a compensated pressure signal sequence;

[0009] Output the compensated pressure signal sequence to a signal processing module of an industrial control system for driving an actuator of the industrial control system.

[0010] In another aspect, an embodiment of the present invention further provides a pressure sensor data accuracy compensation system for industrial control, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, after obtaining the original pressure signal sequence output by the target pressure sensor within a preset acquisition period and the real-time environmental parameter set (covering temperature parameters, humidity parameters and mechanical vibration parameters), the embodiment of the present application introduces a pre-trained compensation parameter generation model, which can generate a dynamic compensation parameter set corresponding to the original pressure signal sequence. Secondly, using the generated dynamic compensation parameter set to gradually adjust each pressure signal in the original pressure signal sequence, deeply exploring the deviation characteristics of each pressure signal under different environmental parameters, can effectively eliminate the pressure signal errors caused by various environmental factors such as temperature, humidity and mechanical vibration, making the compensated pressure signal sequence have extremely high accuracy, greatly improving the reliability and effectiveness of pressure sensor data in industrial control. Finally, outputting the compensated high-precision pressure signal sequence to the signal processing module of the industrial control system to drive the actuator can significantly improve the control accuracy and stability of the industrial control system, making the actions of the actuator more accurate and efficient, avoiding problems such as misoperations and increased defective product rates in the industrial production process caused by inaccurate pressure signals, and thus improving the quality and efficiency of the entire industrial production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a schematic execution flowchart of a pressure sensor data accuracy compensation method for industrial control provided by an embodiment of the present invention.

[0013] Figure 2 is a schematic hardware architecture diagram of a pressure sensor data accuracy compensation system for industrial control provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The present invention will be specifically described below with reference to the accompanying drawings of the specification. Figure 1 is a schematic flowchart of a pressure sensor data accuracy compensation method for industrial control provided by an embodiment of the present invention. The pressure sensor data accuracy compensation method for industrial control will be introduced in detail below.

[0015] Step S110, obtain the original pressure signal sequence output by the target pressure sensor within a preset acquisition period, and the original pressure signal sequence includes a plurality of pressure signals arranged according to time stamps.

[0016] In this embodiment, in an industrial control scenario, taking the reactor in a chemical production workshop as an example, in order to ensure the safety and stability of the reaction process, it is necessary to continuously monitor the pressure in the reactor. The target pressure sensor can be installed at a key part of the reactor. The preset acquisition period can be set to collect data once every 10 seconds. In this way, within a relatively long time period, such as 10 minutes, a number of original pressure signals arranged according to timestamps can be obtained. This original pressure signal reflects the change of the pressure in the reactor over time. For example, at the beginning of the reaction, due to the addition of reactants and the progress of the initial reaction, the pressure may gradually increase. The pressure sensor will record the pressure values at different time points during this increase process, forming an original pressure signal sequence. This original pressure signal contains the real-time state information of the pressure in the reactor.

[0017] Step S120, detect the set of real-time environmental parameters of the target pressure sensor within the acquisition period. The set of real-time environmental parameters includes temperature parameters, humidity parameters, and mechanical vibration parameters.

[0018] The inventors have found through research that for the target pressure sensor installed on the reactor, the environmental parameters around it have an important impact on the accuracy of the pressure signal.

[0019] First of all, during the chemical reaction process, heat is often generated or absorbed, resulting in a change in the temperature around the reactor. The temperature distribution data in the space where the target pressure sensor is located is collected through a temperature sensor array. Suppose one side of the reactor is at a higher temperature because it is close to the heating device, while the other side is relatively lower. The temperature sensor array can detect this temperature difference. Then, spatial weighted average processing is performed on these temperature distribution data to obtain a comprehensive temperature parameter. For example, the sensor data close to the heating device is multiplied by a larger weight because the temperature here may have a greater impact on the pressure sensor, and the sensor data far from the heating device is multiplied by a smaller weight, and finally a parameter that can represent the overall temperature condition around the pressure sensor is obtained.

[0020] In addition, the humidity in the chemical workshop may change due to the water vapor generated in the chemical reaction or the ventilation condition in the workshop. The humidity distribution data in the space where the target pressure sensor is located is collected using a humidity sensor array. There may be some water vapor leakage or condensation phenomena near the reactor, which will cause a relatively high local humidity. After collecting the humidity distribution data, time series filtering processing is performed to remove some abnormal data caused by short-term fluctuations or noise interference of the sensor, and an accurate humidity parameter is obtained.

[0021] In addition, the stirring device in the reactor generates mechanical vibrations during operation, and these mechanical vibrations may be transmitted to the pressure sensor, affecting the accuracy of the pressure signal. The mechanical vibration waveform data at the installation position of the target pressure sensor is collected by a vibration sensor. For example, uneven rotation speed of the stirring device or slight wear of mechanical components may cause irregular changes in the vibration waveform. The frequency-domain energy analysis is performed on the collected mechanical vibration waveform data, and the energy peak value within the preset frequency band is extracted as the mechanical vibration parameter, and this energy peak value can reflect the key characteristics of the influence of mechanical vibration on the pressure sensor. Moreover, the data acquisition timestamps of the temperature parameter, humidity parameter, and mechanical vibration parameter are synchronized and aligned with the timestamp of the original pressure signal sequence, so as to ensure accurate compensation for the pressure signal in the subsequent process.

[0022] Step S130, input the set of real-time environmental parameters into a pre-trained compensation parameter generation model to generate a set of dynamic compensation parameters corresponding to the original pressure signal sequence.

[0023] In the control system of a chemical production workshop, the pre-trained compensation parameter generation model has been trained based on a large amount of experimental data and actual operation data. After obtaining the set of real-time environmental parameters, it is first necessary to perform timestamp alignment verification on the temperature parameter, humidity parameter, and mechanical vibration parameter among them. For example, due to some minor delays or faults in the data acquisition system, the timestamps of individual environmental parameter data may deviate. If the deviation exceeds the preset synchronization threshold, such as 0.1 second, these data need to be excluded, so as to generate a synchronized and calibrated temperature parameter sequence, humidity parameter sequence, and mechanical vibration parameter sequence.

[0024] Then, the synchronized and calibrated temperature parameter sequence is input into the temperature feature extraction channel of the compensation parameter generation model, and this temperature feature extraction channel consists of three layers of convolutional networks. The first layer of convolutional network can capture the low-frequency trend component of the temperature parameter, such as the slow temperature change trend generated by the overall temperature regulation system of the workshop during the long-term operation of the reactor. The second layer of convolutional network can extract the medium-frequency perturbation component of the temperature fluctuation, such as the temperature fluctuation caused by the phased heat release or absorption during the reaction process. The third layer of convolutional network can identify the transient impact component of the temperature mutation, such as the temperature mutation situation when the heating device of the reactor suddenly fails or an emergency cooling operation is carried out. Through the processing of these three layers of convolutional networks, the temperature influence feature vector is finally extracted.

[0025] Next, input the humidity parameter sequence after synchronous calibration into the humidity feature extraction channel of the compensation parameter generation model. This humidity feature extraction channel uses a bidirectional recurrent network to generate humidity influence feature vectors. During the forward propagation process, it can capture the cumulative penetration effect of humidity parameters on the sensor material. For example, over time, a high-humidity environment in a chemical workshop may gradually penetrate into the pressure sensor, affecting its performance. During the backward propagation process, it can identify the hysteresis compensation requirements caused by sudden humidity changes. For example, when the workshop suddenly ventilates and the humidity drops rapidly, the response of the pressure sensor may have a certain hysteresis, and the bidirectional recurrent network can detect this hysteresis phenomenon and generate corresponding humidity influence feature vectors.

[0026] Then, input the mechanical vibration parameter sequence after synchronous calibration into the vibration feature extraction channel of the compensation parameter generation model. The time-frequency hybrid network of this vibration feature extraction channel decomposes the vibration influence feature vector. The time-domain analysis unit extracts the envelope feature of the vibration amplitude. For example, the change trend of the vibration amplitude of a stirring device at different time periods, and this change may have different degrees of influence on the signal of the pressure sensor. The frequency-domain analysis unit separates the distribution ratio of vibration energy at different resonant frequencies. For example, certain specific rotational speeds of a stirring device will excite vibrations at specific frequencies, and the vibration energy distribution at these frequencies is crucial for analyzing the degree of interference on the pressure sensor.

[0027] After that, fuse the temperature influence feature vector, humidity influence feature vector, and vibration influence feature vector, and calculate the coupling strength matrix of the three through the feature cross layer of the compensation parameter generation model. The temperature-humidity coupling strength is characterized by the dot product value of the two-way feature vectors. For example, if the change trends of temperature and humidity are correlated, the dot product value will reflect the comprehensive influence of this correlation on the pressure sensor. The humidity-vibration coupling strength is calculated by the covariance within a sliding window, which can reflect the influence of the synergistic effect between humidity changes and vibration changes on the pressure sensor. The temperature-vibration coupling strength is evaluated using the maximum mutual information entropy, which can reflect the complex mutual relationship between temperature changes and vibration changes.

[0028] Dynamically allocate compensation weight coefficients according to the coupling strength matrix, and generate a real-time proportional combination of temperature compensation weight, humidity compensation weight, and vibration compensation weight through the weight allocator of the compensation parameter generation model. The temperature compensation weight is positively correlated with the temperature-humidity coupling strength, that is, if the coupling between temperature and humidity is stronger, the temperature compensation weight is larger. The humidity compensation weight has an exponential relationship with the humidity-vibration coupling strength, meaning that a small change in the humidity-vibration coupling strength may lead to a large change in the humidity compensation weight. The vibration compensation weight forms a dynamic feedback regulation with the temperature-vibration coupling strength, and dynamically adjusts the vibration compensation weight according to the mutual relationship between the two.

[0029] Finally, input the temperature compensation weight, humidity compensation weight, and vibration compensation weight into the multi-layer compensation parameter generator of the compensation parameter generation model. Generate the basic offset of the temperature compensation coefficient through the first compensation layer, and this basic offset can correct the baseline offset of the pressure sensor caused by temperature changes. The second compensation layer superimposes the sensitivity adjustment amount of the humidity compensation coefficient because humidity may affect the sensitivity of the pressure sensor to pressure changes. The third compensation layer injects the phase correction amount of the vibration compensation coefficient to compensate for the phase change of the pressure signal caused by mechanical vibration. Finally, output a set of dynamic compensation parameters including the temperature compensation coefficient, humidity compensation coefficient, and vibration compensation coefficient. At the same time, it is also necessary to verify the physical feasibility boundary of the set of dynamic compensation parameters, and detect whether each compensation coefficient exceeds the safety threshold of the sensor range through the parameter constraint module of the compensation parameter generation model. If a certain compensation coefficient exceeds the safety threshold, for example, the temperature compensation coefficient is too large, which may cause the compensated pressure signal to exceed the reasonable range, it is necessary to truncate the compensation coefficient that exceeds the safety threshold and generate a compensation parameter correction flag, and match and align the corrected set of dynamic compensation parameters with the time stamps of the original pressure signal sequence to ensure the temporal correspondence between the compensation parameters and the pressure signal.

[0030] Step S140, according to the set of dynamic compensation parameters, perform step-by-step adjustment on each pressure signal in the original pressure signal sequence to obtain a compensated pressure signal sequence.

[0031] In a chemical production workshop, for the original pressure signal sequence obtained from the pressure sensor of the reactor, first perform time-domain filtering processing to obtain a preprocessed pressure signal sequence. For example, due to the presence of various electrical equipment and complex electromagnetic environments in the workshop, the original pressure signal may be interfered by some high-frequency noises. Through time-domain filtering processing, these high-frequency noises can be removed, making the pressure signal smoother and more stable.

[0032] Then, according to the temperature compensation coefficient in the set of dynamic compensation parameters, correct the baseline offset of the preprocessed pressure signal sequence. Assume that due to the increase in the temperature around the reactor, the baseline of the pressure sensor has an upward offset, making the overall pressure signal too high. According to the temperature compensation coefficient, the offset amount that needs to be adjusted can be calculated to restore the baseline of the pressure signal to the normal level.

[0033] Next, according to the humidity compensation coefficient in the set of dynamic compensation parameters, perform sensitivity scaling adjustment on the pressure signal sequence after baseline offset correction. For example, in a high-humidity environment, the sensitivity of the pressure sensor may decrease, resulting in the measured pressure value being smaller than the actual value. Through the humidity compensation coefficient, the pressure signal can be amplified or reduced accordingly to restore its correct sensitivity.

[0034] According to the vibration compensation coefficient in the dynamic compensation parameter set, perform phase synchronization calibration on the pressure signal sequence after sensitivity scaling adjustment. Extract the vibration frequency component corresponding to the mechanical vibration parameter, and generate a reference sine waveform with the same period as the vibration frequency component. For example, if the vibration frequency of the stirring device is 50 Hz, then a 50 Hz reference sine waveform is generated. Perform local extreme point detection on the pressure signal sequence after sensitivity scaling adjustment to determine the phase offset of the signal waveform. Assume that due to the interference of vibration, the phase of the pressure signal has advanced or lagged. Perform phase compensation on the reference sine waveform according to the phase offset and the vibration compensation coefficient in the dynamic compensation parameter set to generate an anti-vibration interference template. Finally, perform a convolution operation on the anti-vibration interference template and the pressure signal sequence after sensitivity scaling adjustment to suppress the signal distortion component caused by vibration, and obtain the compensated pressure signal sequence.

[0035] Step S150, output the compensated pressure signal sequence to the signal processing module of the industrial control system for driving the actuator of the industrial control system.

[0036] In the industrial control system of a chemical production workshop, the compensated pressure signal sequence is output to the signal processing module, which will further analyze and process the pressure signal. For example, compare it with a set pressure threshold to determine whether the pressure in the reactor is within the safe range. If the pressure is too high or too low, the signal processing module will generate corresponding control instructions according to the preset control strategy to drive the actuator to operate. The actuator may be a motor that adjusts the opening degree of the inlet and outlet valves of the reactor, or a controller that controls the power of the heating or cooling device, etc. For example, when the compensated pressure signal indicates that the pressure in the reactor is too high, the signal processing module will send an instruction to the motor that controls the valve opening degree to increase the valve opening degree, thereby reducing the pressure in the reactor and ensuring that the chemical reaction proceeds under safe and stable pressure conditions. The entire process ensures the normal operation of the chemical production workshop and the stability of product quality through precise pressure monitoring and control.

[0037] Another example is the EGR blower throttling system of a certain shipbuilding. The target pressure sensor is installed at a key part of the throttling system to monitor the internal pressure of the system. In the previous steps, the preset acquisition period can be set to collect data every 5 seconds, thus forming the original pressure signal sequence. For example, when the marine engine starts, the EGR blower throttling system starts to work. As factors such as the change in air flow and the operating state of internal components change, the pressure will continuously change. The pressure values at different time points are recorded by the pressure sensor to form the original pressure signal sequence.

[0038] In the above environment, environmental parameters have a certain impact on the pressure sensor. For example, in terms of temperature, since there are many devices running in the ship's engine room, such as the main engine and generator equipment, heat is dissipated, resulting in an uneven distribution of the surrounding environmental temperature. The temperature is higher near these heat-generating devices, which may affect the accuracy of the pressure sensor. Moreover, when the ship is sailing and different speed switches are made, the temperature will also change significantly. For example, when the main engine changes from high-speed operation to low-speed operation, the sharp temperature change will interfere with the normal operation of the pressure sensor.

[0039] Similarly, humidity is also affected by various factors. For example, if the air-conditioning system in the engine room fails or in the humid season, the humidity may increase significantly. For the pressure sensor, high humidity may cause the internal electronic components of the sensor to get damp, thereby affecting the accuracy of the pressure signal. For example, during the rainy season, the humidity in the engine room remains at a high level for a long time, and the pressure signal collected by the pressure sensor may deviate due to the influence of humidity.

[0040] Regarding the mechanical vibration factor, the rotation of the impeller and the operation of the motor inside the EGR blower throttle system itself will generate mechanical vibration, and the vibration of other large equipment in the engine room may also be transmitted through the ground or air. For example, the strong vibration generated by the nearby main engine during operation will be transmitted to the pressure sensor, and these vibrations may cause fluctuations or distortion in the pressure signal collected by the pressure sensor.

[0041] In such a scenario, the compensation parameter generation model combines environmental parameters such as temperature, humidity, and mechanical vibration to generate a set of dynamic compensation parameters to compensate the original pressure signal sequence. The compensated pressure signal sequence is output to the signal processing module of the industrial control system, and this signal processing module can determine whether the pressure of the system is normal based on the compensated pressure signal. If the pressure is too high, for example, due to the blockage of a certain valve in the EGR blower throttle system resulting in an increase in pressure, the signal processing module will send an instruction to the actuator (such as the motor that adjusts the valve opening) to increase the valve opening, thereby reducing the system pressure, ensuring the stable operation of the EGR blower throttle system at an appropriate pressure, and further ensuring the normal operation of the EGR blower throttle system, avoiding problems such as equipment failures or reduced production efficiency caused by pressure problems.

[0042] Based on the above steps, after the embodiment of the present application obtains the original pressure signal sequence output by the target pressure sensor within a preset acquisition period and the real-time environmental parameter set (covering temperature parameters, humidity parameters, and mechanical vibration parameters), a pre-trained compensation parameter generation model is introduced, which can generate a dynamic compensation parameter set corresponding to the original pressure signal sequence. Secondly, using the generated dynamic compensation parameter set to gradually adjust each pressure signal in the original pressure signal sequence, deeply exploring the deviation characteristics of each pressure signal under different environmental parameters, can effectively eliminate the pressure signal errors caused by various environmental factors such as temperature, humidity, and mechanical vibration, making the compensated pressure signal sequence have extremely high accuracy, greatly improving the reliability and effectiveness of the pressure sensor data in industrial control. Finally, outputting the compensated high-precision pressure signal sequence to the signal processing module of the industrial control system to drive the actuator can significantly improve the control accuracy and stability of the industrial control system, making the action of the actuator more accurate and efficient, avoiding problems such as misoperations and increased defective product rates in the industrial production process caused by inaccurate pressure signals, and thus improving the quality and efficiency of the entire industrial production process.

[0043] In a possible implementation manner, step S120 includes:

[0044] Step S121, collecting temperature distribution data in the space where the target pressure sensor is located through a temperature sensor array, and performing spatial weighted average processing on the temperature distribution data to obtain the temperature parameter.

[0045] In this embodiment, the temperature environment around the reactor has an important influence on the measurement accuracy of the pressure sensor. A temperature sensor array is arranged to collect the temperature distribution data in the space where the target pressure sensor is located. Since the temperatures in different parts of the reactor may vary, for example, the temperature in the area near the heating jacket of the reactor is relatively high, while the temperature in the area far from the heat source and near the workshop ventilation opening is relatively low. Each sensor in the temperature sensor array is distributed at different positions around the target pressure sensor to comprehensively obtain temperature information. The above sensors transmit the temperature data they detect back to form temperature distribution data. Then, spatial weighted averaging is performed on these temperature distribution data to obtain a temperature parameter. In this process, the weights are determined according to the relationship between the position of each sensor and the target pressure sensor and the degree of influence of the temperature at that position on the pressure sensor. For example, the sensor data in the area closer to the target pressure sensor and with a greater impact of temperature change on the pressure sensor is given a larger weight, while the sensor data in the area farther away and with a relatively smaller impact is given a smaller weight. The temperature parameter obtained after weighted average calculation can accurately reflect the overall temperature condition around the target pressure sensor. This temperature parameter is closely related to the working state of the pressure sensor and is an important basis for subsequent precise compensation of the pressure signal.

[0046] Step S122: Collect humidity distribution data in the space where the target pressure sensor is located through a humidity sensor array, and perform time series filtering on the humidity distribution data to obtain the humidity parameter.

[0047] The humidity situation in a chemical workshop is relatively complex, and changes in humidity can affect the performance of the pressure sensor. A humidity sensor array is used to collect humidity distribution data in the space where the target pressure sensor is located. Around the reactor, the humidity at different positions may vary due to the evaporation of water vapor during the chemical reaction process or the ventilation conditions in the workshop. For example, near the material inlet of the reactor, the humidity may be relatively high due to the moisture carried by the material or the water vapor generated at the beginning of the reaction; while the humidity is relatively low in the area with better air circulation in the workshop. The humidity distribution data detected by the humidity sensor array may contain some noise data caused by the short-term fluctuations of the sensors themselves or minor local environmental interferences. To obtain an accurate humidity parameter, time series filtering needs to be performed on the humidity distribution data. This filtering process can remove the abnormal data that does not conform to the humidity change trend, thereby obtaining a humidity parameter that can truly reflect the humidity condition around the target pressure sensor. This humidity parameter is also indispensable for accurately compensating the pressure signal of the pressure sensor.

[0048] Step S123, collecting mechanical vibration waveform data of the target pressure sensor installation position through a vibration sensor, and performing frequency domain energy analysis on the mechanical vibration waveform data, extracting the energy peak in a preset frequency band as the mechanical vibration parameter. The data collection timestamps of the temperature parameter, humidity parameter and mechanical vibration parameter are synchronized with the timestamp of the original pressure signal sequence.

[0049] The stirring device in the reactor will generate mechanical vibration during operation, which will be transmitted to the target pressure sensor installed on the reactor, thereby affecting the accuracy of the pressure signal. The mechanical vibration waveform data of the target pressure sensor installation position is collected by the vibration sensor. Due to the influence of factors such as the mechanical structure, rotation speed and load of the stirring device, the mechanical vibration waveform data has complex characteristics. For example, the wear and imbalance of the stirring blades or the uneven distribution of materials in the reactor may cause changes in the vibration waveform. The collected mechanical vibration waveform data is subjected to frequency domain energy analysis. In this process, the vibration waveform is converted from the time domain to the frequency domain to analyze the energy distribution under different frequency components. By extracting the energy peak in the preset frequency band as the mechanical vibration parameter, the frequency band where the energy peak is located may be a frequency band related to the natural frequency of the stirring device or the key vibration frequency during its operation. Extracting this energy peak can highlight the key features of the impact of mechanical vibration on the pressure sensor, thereby providing an important basis for accurately considering the impact of mechanical vibration in the subsequent compensation parameter generation model. In addition, during the entire acquisition process, the data acquisition timestamps of the temperature parameters, humidity parameters and mechanical vibration parameters are synchronized and aligned with the timestamps of the original pressure signal sequence. This is a very critical point, because only by aligning the timestamps synchronously can we ensure that in the subsequent compensation process, the environmental parameters are accurately associated with the corresponding pressure signals, thereby achieving accurate compensation for the pressure signal based on environmental influences, ensuring the accuracy and reliability of pressure monitoring in the chemical reactor, and thus ensuring the stability and safety of the entire chemical production process.

[0050] In a possible implementation, step S130 includes:

[0051] Step S131, performing timestamp alignment verification on the temperature parameters, humidity parameters and mechanical vibration parameters in the real-time environmental parameter set, eliminating environmental parameter data whose timestamp deviation exceeds a preset synchronization threshold, and generating a temperature parameter sequence, humidity parameter sequence and mechanical vibration parameter sequence after synchronization calibration.

[0052] In the actual scenario of chemical production, due to the complexity of the data acquisition system, there may be a deviation in the timestamps of data collected by different sensors. For example, when the temperature sensor array, humidity sensor array, and vibration sensor collect data, due to their respective acquisition frequencies, transmission delays, or slight differences in internal clocks, the timestamps of the collected temperature parameters, humidity parameters, and mechanical vibration parameters may not be exactly the same. The preset synchronization threshold is set according to the system's requirements for data synchronization, such as being set to 0.05 seconds. When it is found that the timestamp deviation of a certain environmental parameter data exceeds this preset synchronization threshold, it needs to be excluded. Through such processing, it is possible to generate temperature parameter sequences, humidity parameter sequences, and mechanical vibration parameter sequences after synchronous calibration, ensuring that the data for subsequent processing has strict consistency in time.

[0053] Step S132, input the temperature parameter sequence after synchronous calibration into the temperature feature extraction channel of the compensation parameter generation model, and extract the temperature influence feature vector through the three-layer convolutional network of the temperature feature extraction channel. Among them, the first-layer convolutional network captures the low-frequency trend component of the temperature parameter, the second-layer convolutional network extracts the intermediate-frequency perturbation component of the temperature fluctuation, and the third-layer convolutional network identifies the transient impact component of the temperature mutation.

[0054] In this embodiment, the temperature feature extraction channel includes a three-layer convolutional network, and each layer of the convolutional network has its specific function. The first layer of the convolutional network is used to capture the low-frequency trend component of the temperature parameter. During the operation of the chemical reaction kettle, the temperature around the reaction kettle will be affected by the overall temperature regulation system of the workshop, resulting in a slow change trend. For example, in order to maintain the overall temperature stability, the cooling system of the workshop will continuously perform refrigeration regulation on the workshop environment, which makes the temperature around the reaction kettle show a slow downward trend over a long period of time. This low-frequency trend component reflects the basic change law of the temperature over a long period of time and is crucial for accurately evaluating the impact of temperature on the pressure sensor subsequently. The second layer of the convolutional network is used to extract the medium-frequency disturbance component of the temperature fluctuation. During the reaction process, the chemical reaction in the reaction kettle will periodically release or absorb heat, resulting in fluctuations in the temperature around the reaction kettle. For example, during the exothermic stage of a certain chemical reaction, the temperature around the reaction kettle will rise rapidly and then gradually decrease when the reaction tends to be stable. This temperature fluctuation is the medium-frequency disturbance component. The second layer of the convolutional network can accurately capture this fluctuation situation and provide a basis for analyzing the dynamic impact of temperature on the pressure sensor. The third layer of the convolutional network is responsible for identifying the transient shock component of the temperature mutation. In chemical production, some sudden situations may occur, such as the failure of the heating device or an emergency cooling operation. For example, the heating element of the heating device suddenly short-circuits, resulting in an instantaneous decrease in the heating power, and the temperature around the reaction kettle will change suddenly. The third layer of the convolutional network can accurately identify this transient shock component of the temperature mutation, comprehensively obtain various influence characteristics of the temperature parameter on the pressure sensor, and finally extract the temperature influence feature vector.

[0055] Step S133: Input the humidity parameter sequence after synchronous calibration into the humidity feature extraction channel of the compensation parameter generation model, and generate a humidity influence feature vector through the bidirectional recurrent network in the humidity feature extraction channel, where the forward propagation process captures the cumulative penetration effect of the humidity parameter on the sensor material, and the backward propagation process identifies the hysteresis compensation requirement caused by the humidity mutation.

[0056] In a chemical workshop, in an environment with a relatively high humidity, over time, water vapor will gradually penetrate into the internal materials of the pressure sensor. For example, there may be tiny gaps in the sensor's housing, and water vapor will slowly enter the sensor through these gaps, affecting the electrical or mechanical properties of the sensor. The bidirectional recurrent network for forward propagation can accurately capture this cumulative penetration effect based on the changing trend of humidity parameters. During the backpropagation process, the bidirectional recurrent network can identify the need for hysteresis compensation caused by sudden changes in humidity. For example, when the workshop suddenly conducts ventilation operations, the humidity in the workshop will drop rapidly. However, due to its own thermal inertia or the moisture absorption-desorption characteristics of the materials, the pressure sensor may have a certain lag in its response to humidity changes. The backpropagation process can detect this lag phenomenon and generate a corresponding humidity influence eigenvector, thus providing accurate humidity-related information for subsequent compensation processing.

[0057] Step S134: Input the synchronously calibrated mechanical vibration parameter sequence into the vibration feature extraction channel of the compensation parameter generation model, and decompose the vibration influence eigenvector through the time-frequency hybrid network of the vibration feature extraction channel, where the time-domain analysis unit extracts the envelope feature of the vibration amplitude, and the frequency-domain analysis unit separates the distribution ratio of the vibration energy at different resonant frequencies.

[0058] In a chemical reactor, the operating state of the stirring device will directly affect the mechanical vibration situation. For example, during long-term operation, the stirring blades may experience wear or imbalance, which will cause changes in the vibration amplitude. The time-domain analysis unit can extract the changing trend of the vibration amplitude in different time periods to form the envelope feature of the vibration amplitude, which reflects the overall changing law of mechanical vibration in the time domain and is of great significance for analyzing the impact of vibration on the pressure sensor. The frequency-domain analysis unit separates the distribution ratio of the vibration energy at different resonant frequencies. During the operation of the stirring device, due to its own mechanical structure and working principle, it will generate relatively large vibration energy at specific resonant frequencies. For example, when the rotation speed of the stirring device reaches a certain specific value, it will excite a certain resonant frequency of the reactor structure, resulting in a significant increase in the vibration energy at this frequency. The frequency-domain analysis unit can accurately separate the distribution ratio of the vibration energy at different resonant frequencies, thereby comprehensively obtaining the influence characteristics of mechanical vibration on the pressure sensor.

[0059] Step S135: Fuse the temperature influence eigenvector, humidity influence eigenvector, and vibration influence eigenvector, and calculate the coupling strength matrix of the three through the feature cross layer of the compensation parameter generation model, where the temperature-humidity coupling strength is characterized by the dot product value of the two-way eigenvector, the humidity-vibration coupling strength is calculated through the covariance within the sliding window, and the temperature-vibration coupling strength is evaluated using the maximum mutual information entropy.

[0060] For example, for the intensity of the temperature-humidity coupling effect, it is characterized by the dot product value of the two-way eigenvector. In the environment of a chemical workshop, there may be complex interrelationships between temperature and humidity. For example, when the temperature is high, the evaporation rate of water vapor accelerates, which may lead to a decrease in humidity in the workshop; conversely, when the temperature is low, water vapor may condense more easily, resulting in an increase in humidity. The mutual correlation between temperature and humidity does not simply superimpose on the pressure sensor, but accurately reflects the comprehensive influence of the coupling effect of the two on the pressure sensor through the dot product value. The intensity of the humidity-vibration coupling effect is calculated by the covariance within a sliding window. In actual situations, changes in humidity may affect the friction coefficient between mechanical components, thereby affecting the vibration characteristics of the stirring device. For example, in a high-humidity environment, mechanical components may rust or become more lubricated on the surface, which will change the vibration of the stirring device. By calculating the covariance within the sliding window, the intensity of the humidity-vibration coupling effect can be accurately measured, reflecting the degree of influence of this coupling effect on the pressure sensor. The intensity of the temperature-vibration coupling effect is evaluated using the maximum mutual information entropy. In a chemical reactor system, changes in temperature may cause thermal expansion or contraction of the reactor material, thereby changing the mechanical structure characteristics of the reactor and further affecting the vibration of the stirring device. The maximum mutual information entropy can accurately evaluate the intensity of the temperature-vibration coupling effect from complex relationships, providing a basis for comprehensively analyzing the comprehensive influence of environmental parameters on the pressure sensor.

[0061] Step S136, dynamically allocate compensation weight coefficients according to the coupling effect intensity matrix, and generate a real-time proportional combination of temperature compensation weight, humidity compensation weight, and vibration compensation weight through the weight allocator of the compensation parameter generation model, where the temperature compensation weight is positively correlated with the intensity of the temperature-humidity coupling effect, the humidity compensation weight has an exponential relationship with the intensity of the humidity-vibration coupling effect, and the vibration compensation weight forms a dynamic feedback regulation with the intensity of the temperature-vibration coupling effect.

[0062] The temperature compensation weight is positively correlated with the intensity of the temperature-humidity coupling effect. This means that when the intensity of the temperature-humidity coupling effect is large, it indicates that the mutual influence between temperature and humidity has a greater comprehensive impact on the pressure sensor, and a larger weight for temperature compensation needs to be given. For example, if the change trends of temperature and humidity in the workshop are closely correlated and have a significant impact on the measurement accuracy of the pressure sensor, then the temperature compensation weight will increase accordingly. The humidity compensation weight has an exponential relationship with the intensity of the humidity-vibration coupling effect. In actual scenarios, a slight change in the intensity of the humidity-vibration coupling effect may have a greater impact on the pressure sensor. For example, when the intensity of the humidity-vibration coupling effect slightly increases, due to the exponential relationship, the humidity compensation weight will increase significantly to adapt to the impact of this coupling effect on the pressure sensor. The vibration compensation weight forms a dynamic feedback regulation with the intensity of the temperature-vibration coupling effect. When the intensity of the temperature-vibration coupling effect changes, the vibration compensation weight will be dynamically adjusted according to this change. For example, as the temperature of the reaction kettle changes, resulting in a change in the vibration characteristics of the stirring device, the vibration compensation weight will be adjusted accordingly to accurately compensate for the impact of vibration on the pressure sensor.

[0063] Step S137, input the temperature compensation weight, humidity compensation weight, and vibration compensation weight into the multi-layer compensation parameter generator of the compensation parameter generation model, generate the basic offset of the temperature compensation coefficient through the first compensation layer, superimpose the sensitivity adjustment amount of the humidity compensation coefficient through the second compensation layer, inject the phase correction amount of the vibration compensation coefficient through the third compensation layer, and finally output a set of dynamic compensation parameters including the temperature compensation coefficient, humidity compensation coefficient, and vibration compensation coefficient.

[0064] In the environment of a chemical reaction kettle, due to the influence of temperature, the measurement baseline of the pressure sensor may shift. For example, when the temperature rises, the electrical characteristics of the pressure sensor may change, resulting in a fixed offset in the measured pressure value. The basic offset of the temperature compensation coefficient generated by the first compensation layer is to correct this baseline shift caused by temperature changes. The second compensation layer superimposes the sensitivity adjustment amount of the humidity compensation coefficient. In a high-humidity environment, the sensitivity of the pressure sensor may decrease, that is, the response ability to pressure changes weakens. The sensitivity adjustment amount of the humidity compensation coefficient can adjust the pressure signal accordingly according to the degree of influence of humidity on the sensor sensitivity to restore the normal sensitivity of the sensor. The third compensation layer injects the phase correction amount of the vibration compensation coefficient. The vibration of the stirring device will cause a change in the phase of the pressure signal. For example, the vibration may cause the waveform of the pressure signal to shift on the time axis. The phase correction amount of the vibration compensation coefficient injected by the third compensation layer is to compensate for this phase change caused by vibration. Finally, a set of dynamic compensation parameters including the temperature compensation coefficient, humidity compensation coefficient, and vibration compensation coefficient is output.

[0065] Step S138: Verify the physical feasibility boundary of the dynamic compensation parameter set. Use the parameter constraint module of the compensation parameter generation model to detect whether each compensation coefficient exceeds the safety threshold of the sensor range. Truncate the compensation coefficients that exceed the safety threshold and generate a compensation parameter correction flag. Match and align the corrected dynamic compensation parameter set with the timestamps of the original pressure signal sequence.

[0066] In chemical production, a pressure sensor has a specified measurement range. If the compensation coefficient is too large, it may cause the compensated pressure signal to exceed the sensor range, resulting in incorrect measurement results. For example, if the temperature compensation coefficient is too large, it may cause the pressure signal that was originally within the range to exceed the upper limit of the range after compensation. When it is detected that a certain compensation coefficient exceeds the safety threshold, the excess part needs to be truncated and a compensation parameter correction flag is generated. Then, match and align the corrected dynamic compensation parameter set with the timestamps of the original pressure signal sequence to ensure the accurate correspondence between the compensation parameters and the original pressure signal in terms of time, thereby providing a reliable basis for accurately compensating the original pressure signal sequence according to the dynamic compensation parameter set and ensuring the accuracy and reliability of pressure monitoring and control in the chemical reactor.

[0067] In a possible implementation, the training steps of the compensation parameter generation model include:

[0068] Step S210: Under the conditions of multiple different combinations of environmental parameters, collect the pressure signal sample set and the corresponding standard pressure signal set of the target pressure sensor.

[0069] The environment of a chemical production workshop is complex and changeable, and different combinations of environmental parameters will have different effects on the measurement of pressure sensors. For example, at different operating stages of a reaction kettle, factors such as the intensity of the reaction, the heating or cooling rate, and the flow rate of the material will cause changes in the temperature, humidity, and mechanical vibration of the surrounding environment. To comprehensively cover all possible situations, data collection needs to be carried out under various combinations of environmental parameters. For the pressure signal sample set, this is the pressure signal inside the reaction kettle measured by the target pressure sensor under various environmental conditions during the actual chemical production process. The standard pressure signal set, on the other hand, is the signal corresponding to the true pressure value inside the reaction kettle obtained through more accurate and reliable measurement means or measurement equipment that has been strictly calibrated. For example, a high-precision pressure measurement instrument can be used, whose accuracy is much higher than that of the target pressure sensor and has been professionally calibrated and verified to provide accurate pressure standard values, which are used as the standard pressure signal set. The collection of these data needs to be carried out over a long time period to ensure that enough data samples of different situations can be obtained, thus providing a rich data basis for subsequent model training.

[0070] Step S220, for each combination of environmental parameters, calculate the error distribution matrix between the pressure signal sample set and the standard pressure signal set.

[0071] In the actual scenario of chemical production, due to the influence of environmental factors on the target pressure sensor, there will be an error between its measured value and the standard value. For each specific combination of environmental parameters, each sample in the pressure signal sample set is compared with the sample corresponding to the same timestamp in the standard pressure signal set, and the error value between them is calculated. For example, at a certain moment, when the environmental temperature is high, the humidity is moderate, and the mechanical vibration is small, the difference between the pressure value measured by the target pressure sensor and the pressure value obtained by the standard pressure measurement instrument is an error value. Arranging all these error values in chronological order and according to the corresponding combination of environmental parameters forms the error distribution matrix, which details the distribution of pressure signal measurement errors under different combinations of environmental parameters and provides a key basis for determining the parameter mapping relationship in the compensation parameter generation model.

[0072] Step S230, perform non-linear regression fitting on the combination of environmental parameters and the error distribution matrix to determine the parameter mapping relationship in the compensation parameter generation model.

[0073] Step S240, iteratively optimize the weight coefficients in the parameter mapping relationship through the backpropagation algorithm until the matching degree between the predicted compensation parameters output by the compensation parameter generation model and the error distribution matrix reaches a preset threshold.

[0074] In a possible implementation, step S230 includes:

[0075] Step S231, performing three-dimensional surface fitting on the temperature parameter, humidity parameter, and mechanical vibration parameter in the environmental parameter combination respectively with the error amplitude corresponding to the time stamp in the error distribution matrix, to generate an initial parameter mapping relationship including a temperature compensation surface, a humidity compensation surface, and a vibration compensation surface.

[0076] In the environment of a reaction kettle in a chemical workshop, the influence of temperature, humidity, and mechanical vibration on the pressure sensor is non-linear. For example, the influence of temperature on the pressure sensor may show a certain change trend within a certain temperature range, while having a different change trend within another temperature range. Through three-dimensional surface fitting, the complex relationship between the temperature parameter and the corresponding error amplitude can be represented by the temperature compensation surface. Similarly, the humidity compensation surface and the vibration compensation surface also respectively reflect the relationships between the humidity parameter, the mechanical vibration parameter, and the error amplitude. These surfaces together constitute the initial parameter mapping relationship, which preliminarily describes the mapping relationship between the environmental parameters and the pressure signal error.

[0077] Step S232, calculating the error change rate of each pressure signal sample within a continuous acquisition period, to generate a dynamic gradient vector of the error distribution matrix.

[0078] During the chemical production process, the error of the pressure sensor is not static. As the operation time of the reaction kettle progresses, due to the continuous action of environmental factors and the characteristic changes of the sensor itself, the error will change dynamically. For example, as the reaction proceeds, the temperature inside the reaction kettle gradually rises, the humidity may increase due to the water vapor generated by the chemical reaction, and the mechanical vibration may also change due to the wear of the stirring device. These factors will cause the error of the pressure sensor to have different change rates in different acquisition periods. Arranging the error change rates of each pressure signal sample within a continuous acquisition period in a certain order forms the dynamic gradient vector of the error distribution matrix. This dynamic gradient vector reflects the trend of the error changing with time and environmental factors, providing important dynamic information for subsequent multivariable coupling analysis.

[0079] Step S233, performing multivariable coupling analysis on the temperature parameter change rate, humidity parameter change rate, and mechanical vibration parameter change rate in the environmental parameter combination with the dynamic gradient vector, to generate a set of dynamic compensation coefficients including a temperature gradient weight, a humidity gradient weight, and a vibration gradient weight.

[0080] In a chemical reaction kettle system, there are complex interrelationships among temperature, humidity, and mechanical vibration, and the influence of their change rates on the error of the pressure sensor is not independent. For example, a rapid increase in temperature may cause the humidity to change more quickly and may also affect the mechanical vibration characteristics of the stirring device, thereby jointly affecting the error change rate of the pressure sensor. Through multivariable coupling analysis, the comprehensive influence degree of the change rates of temperature, humidity, and mechanical vibration parameters on the error change rate can be accurately measured, thereby generating a temperature gradient weight, a humidity gradient weight, and a vibration gradient weight. These weights reflect the relative importance of each environmental parameter to the error change of the pressure sensor under different environmental parameter change conditions and constitute a set of dynamic compensation coefficients.

[0081] Step S234: Perform a point-by-point multiplication operation on the temperature compensation surface, humidity compensation surface, and vibration compensation surface in the initial parameter mapping relationship with the corresponding gradient weights in the set of dynamic compensation coefficients respectively to generate an enhanced parameter mapping relationship that combines static error characteristics and dynamic error characteristics.

[0082] The error of the pressure sensor includes both static error caused by the static values of environmental parameters and dynamic error caused by the changes in environmental parameters. For example, the basic temperature level around the reaction kettle will cause a certain static error in the pressure sensor, and the change rate of temperature will introduce dynamic error. Through the above point-by-point multiplication operation, the part reflecting the static error characteristics in the temperature compensation surface, humidity compensation surface, and vibration compensation surface is combined with the gradient weights reflecting the dynamic error characteristics in the set of dynamic compensation coefficients, so that the enhanced parameter mapping relationship can more comprehensively and accurately describe the relationship between environmental parameters and the error of the pressure sensor, taking into account both the static influence of environmental parameters and the influence of their dynamic changes.

[0083] Step S235: Construct a neural network model with a multi-layer hidden structure, input the temperature compensation surface data, humidity compensation surface data, and vibration compensation surface data in the enhanced parameter mapping relationship into three independent input channels of the neural network model respectively, and perform non-linear feature fusion on the data of the three input channels through the fully connected layer of the neural network model to generate a final parameter mapping relationship with a cross-compensation effect.

[0084] The effects of temperature, humidity, and mechanical vibration on the pressure sensor are not simply linearly superimposed, but there are complex cross - influence relationships. For example, temperature and humidity may act together to affect the electrical performance of the pressure sensor, and humidity and mechanical vibration may jointly change the mechanical structure stability of the pressure sensor, thereby affecting its measurement accuracy. The multi - layer hidden structure of the neural network model has a powerful non - linear fitting ability and can learn this complex cross - influence relationship. By inputting the temperature compensation surface data, humidity compensation surface data, and vibration compensation surface data into three independent input channels respectively, the fully - connected layer can perform non - linear feature fusion on these data, comprehensively process the data from different channels, so that the finally generated final parameter mapping relationship can reflect the cross - compensation effect among temperature, humidity, and mechanical vibration, and more accurately reflect the complex relationship between environmental parameters and the error of the pressure sensor.

[0085] Step S236: Optimize the output accuracy of the final parameter mapping relationship. By performing residual backpropagation calculation on the output layer data of the neural network model and the standard pressure signal set, adjust the node connection weights of the fully - connected layer in the neural network model until the prediction error value of the final parameter mapping relationship reaches a preset accuracy threshold.

[0086] To ensure that the compensation parameter generation model can accurately predict compensation parameters, it is necessary to optimize the output accuracy of the final parameter mapping relationship. Compare the output layer data of the neural network model with the standard pressure signal set, calculate the residual between the two, and this residual reflects the gap between the prediction result of the current final parameter mapping relationship and the true standard value. Through residual backpropagation calculation, according to the magnitude and direction of the residual, adjust the node connection weights of the fully - connected layer in the neural network model. For example, if the residual between the predicted pressure signal and the standard pressure signal is large under a certain combination of environmental parameters, it indicates that there is a deviation in the current parameter mapping relationship, and the node connection weights need to be adjusted to reduce this deviation. Continuously repeat this process, gradually adjust the node connection weights until the prediction error value of the final parameter mapping relationship reaches the preset accuracy threshold, ensuring that the final parameter mapping relationship can accurately reflect the relationship between environmental parameters and the error of the pressure sensor, thereby providing a reliable basis for generating accurate compensation parameters.

[0087] Step S237: Solidify the optimized node connection weights in the neural network model to generate the data structure of the executable parameter mapping relationship in the compensation parameter generation model, where the data structure includes cross - query interfaces for the temperature compensation surface index table, humidity compensation surface index table, and vibration compensation surface index table.

[0088] Specifically, once the node connection weights of the neural network model are optimized to meet the preset accuracy requirements, these weights need to be solidified to form a data structure that generates an executable parameter mapping relationship in the compensation parameter model. The cross-query interfaces of the temperature compensation surface index table, humidity compensation surface index table, and vibration compensation surface index table in this data structure are of great significance. For example, during the actual pressure signal compensation process, when it is necessary to find the corresponding compensation parameters according to the current environmental parameter combination, the cross-query interfaces of these index tables can be used to quickly and accurately obtain the compensation surface information corresponding to temperature, humidity, and mechanical vibration, and then determine the appropriate compensation parameters to achieve accurate compensation of the pressure sensor signal and ensure the accuracy and reliability of pressure monitoring in the chemical reactor.

[0089] In a possible implementation manner, step S140 includes:

[0090] Step S141, performing time-domain filtering processing on the original pressure signal sequence to obtain a preprocessed pressure signal sequence.

[0091] For example, the original pressure signal sequence is collected by a pressure sensor installed at a key part of the reactor. However, this original signal sequence may be affected by various interference factors. For example, there are various electrical devices in the workshop, and these devices may generate electromagnetic interference during operation, resulting in high-frequency noise components mixed in the pressure signal. In addition, due to the start-stop operations of various devices or the instability of material flow during the chemical production process, some irregular fluctuations may also be introduced into the pressure signal. Through time-domain filtering processing, these interference components can be effectively removed. Time-domain filtering technology can identify and filter out high-frequency noise and irregular fluctuations that do not conform to the normal change law of the pressure signal according to the characteristics of the signal in the time domain. For example, a low-pass filter is used, which allows pressure signal components below a certain cut-off frequency to pass through, while noise and interference components above this cut-off frequency are filtered out. After such time-domain filtering processing, the obtained preprocessed pressure signal sequence is smoother and more stable, and can more accurately reflect the actual pressure change situation in the reactor, providing a more reliable basis for subsequent adjustment operations.

[0092] Step S142, performing baseline offset correction on the preprocessed pressure signal sequence according to the temperature compensation coefficient in the dynamic compensation parameter set.

[0093] In the environment of a chemical workshop, temperature has an important impact on pressure sensors. During the operation of a reactor, due to the exothermic or endothermic process of internal chemical reactions and the fluctuations in the surrounding environmental temperature, the temperature around the pressure sensor will change. For example, during some stages with intense exothermic reactions, the temperature around the reactor may increase significantly. The change in temperature will cause changes in the electrical or mechanical characteristics of the pressure sensor, which in turn leads to a baseline shift in the pressure signal. This baseline shift is manifested as a certain value of upward or downward translation of the entire pressure signal. The temperature compensation coefficient in the dynamic compensation parameter set is obtained based on the previous precise modeling and analysis of the relationship between temperature and the pressure sensor. Through this temperature compensation coefficient, the amount of baseline shift correction required for the pressure signal can be accurately calculated. For example, if the increase in temperature causes the baseline of the pressure signal to shift upward by a certain value, a reverse adjustment amount can be determined according to the temperature compensation coefficient to restore the baseline of the pressure signal to the normal level, enabling the pressure signal to accurately reflect the actual pressure inside the reactor and avoiding pressure measurement errors caused by baseline shift due to temperature.

[0094] Step S143: According to the humidity compensation coefficient in the dynamic compensation parameter set, perform sensitivity scaling adjustment on the pressure signal sequence after baseline shift correction.

[0095] The humidity environment in a chemical workshop is relatively complex, and the change in humidity will affect the sensitivity of the pressure sensor. During the operation of the reactor, due to the possible generation of water vapor during chemical reactions or changes in the ventilation conditions in the workshop, the humidity of the environment where the pressure sensor is located will change. For example, when the ventilation in the workshop is poor, the humidity around the reactor may increase. The increase in humidity may cause some sensitive components inside the pressure sensor to be affected by moisture, thereby reducing its sensitivity to pressure changes. This means that under the same pressure change, the amplitude of the signal output by the pressure sensor may become smaller. The humidity compensation coefficient in the dynamic compensation parameter set reflects the degree of influence of humidity on the sensitivity of the pressure sensor. According to this humidity compensation coefficient, sensitivity scaling adjustment can be performed on the pressure signal sequence after baseline shift correction. If humidity causes a decrease in sensor sensitivity, an amplification factor can be calculated through the humidity compensation coefficient to amplify the amplitude of the pressure signal accordingly so that it can accurately reflect the actual pressure change. Conversely, if the change in humidity causes an increase in sensitivity, a corresponding reduction adjustment is made to ensure that the sensitivity of the pressure signal matches the actual pressure change situation, thereby improving the accuracy of pressure measurement.

[0096] Step S144: According to the vibration compensation coefficient in the dynamic compensation parameter set, perform phase synchronization calibration on the pressure signal sequence after sensitivity scaling adjustment.

[0097] Step S145, using the signal sequence after phase synchronization calibration as the compensated pressure signal sequence.

[0098] Among them, step S144 includes:

[0099] Step S1441, extracting the vibration frequency component corresponding to the mechanical vibration parameter and generating a reference sine waveform with the same period as the vibration frequency component.

[0100] In a chemical reactor, the operation of the stirring device will generate mechanical vibrations, which will be transmitted to the pressure sensor and affect the phase of the pressure signal. The mechanical vibration parameter reflects the characteristics of this vibration. By performing frequency domain analysis on the mechanical vibration parameter, the vibration frequency component can be accurately extracted. For example, if the operating frequency of the stirring device is 50 Hz, then the corresponding vibration frequency component is 50 Hz. Based on this vibration frequency component, a reference sine waveform with the same period can be generated. This reference sine waveform has the same frequency characteristics as the vibration, providing a benchmark for subsequent phase compensation.

[0101] Step S1442, performing local extreme point detection on the pressure signal sequence after sensitivity scaling adjustment to determine the phase offset of the signal waveform.

[0102] Due to the interference of mechanical vibrations, the phase of the pressure signal may shift. By detecting the local extreme points in the pressure signal waveform, such as the positions of the peaks and valleys, the phase shift of the pressure signal relative to the normal situation can be analyzed. For example, in the absence of vibration interference, the peak of the pressure signal should appear at a specific time point, but due to the influence of vibration, the peak may appear earlier or later. The amount of this advance or delay reflects the phase offset of the signal waveform.

[0103] Step S1443, performing phase compensation on the reference sine waveform according to the phase offset and the vibration compensation coefficient in the dynamic compensation parameter set to generate an anti-vibration interference template.

[0104] The vibration compensation coefficient is obtained based on the previous modeling and analysis of the relationship between vibration and the pressure sensor. It reflects the degree of compensation required for the phase shift caused by vibration. According to the magnitude and direction of the phase offset, as well as the vibration compensation coefficient, precise phase adjustment can be performed on the reference sine waveform. For example, if the phase offset is positive, indicating that the pressure signal phase is advanced, then according to the vibration compensation coefficient, the phase of the reference sine waveform can be adjusted backward by a certain amount to generate an anti-vibration interference template. This anti-vibration interference template has the characteristics that match the phase characteristics that the pressure signal should have without vibration interference.

[0105] Step S1444, perform a convolution operation on the anti-vibration interference template and the pressure signal sequence after sensitivity scaling adjustment to suppress the signal distortion components caused by vibration.

[0106] Convolution operation is an effective signal processing method that can incorporate the characteristics of the anti-vibration interference template into the pressure signal sequence. In the environment of a chemical reaction kettle, due to the interference of vibration, the pressure signal may exhibit waveform distortion, such as changes in the shape of wave peaks and valleys, or unstable fluctuations in the signal amplitude over different periods. Through the convolution operation, the anti-vibration interference template can suppress these distortion components in the pressure signal sequence. For example, when the anti-vibration interference template is convolved with the pressure signal sequence, it will perform weighted averaging or correction on the distorted parts of the pressure signal according to its own phase and amplitude characteristics, making the finally obtained signal sequence closer to the ideal pressure signal without vibration interference, thereby improving the accuracy and reliability of the pressure signal. The signal sequence after phase synchronization calibration is used as the compensated pressure signal sequence, which can more accurately reflect the actual pressure situation in the reaction kettle and provide reliable data support for pressure monitoring and control in the chemical production process.

[0107] In a possible implementation manner, the method further includes:

[0108] Step S310, monitor the change gradient of the environmental parameters of the target pressure sensor within consecutive acquisition periods.

[0109] For example, the environmental parameters around the reaction kettle, such as temperature, humidity, and mechanical vibration, are constantly in dynamic change. Taking temperature as an example, the chemical reaction in the reaction kettle will continuously release or absorb heat, which will cause the temperature around the reaction kettle to change. Within consecutive acquisition periods, it may be found that the temperature slowly rises in some periods and then drops in the following periods. This rate of change of temperature over time is the temperature change gradient. Similarly, humidity will also change due to factors such as the generation of water vapor during the reaction process and the operation of the workshop ventilation system, and its change situation within consecutive acquisition periods constitutes the humidity change gradient. In terms of mechanical vibration, the operating state of the stirring device, the wear of mechanical components, and the stirring resistance of the material will all change the mechanical vibration situation, and these changes form the mechanical vibration change gradient within consecutive acquisition periods.

[0110] Step S320, when the change gradient of the environmental parameters exceeds the preset fluctuation threshold, re-acquire the current set of environmental parameters and input them into the compensation parameter generation model to generate an updated set of dynamic compensation parameters.

[0111] For example, the fluctuation threshold of the preset temperature is 5°C change every 10 minutes. If the temperature change gradient is found to exceed this threshold within a certain continuous acquisition period, this may indicate that some abnormal conditions have occurred inside the reactor, such as a runaway reaction or a heating-cooling system failure. At this time, it is necessary to immediately re-acquire the current set of environmental parameters including temperature, humidity, and mechanical vibration. For the re-acquisition of temperature, to ensure the accuracy and timeliness of the acquisition, it may be necessary to increase the acquisition frequency of the temperature sensor or use a more precise temperature acquisition device. For the acquisition of humidity, considering the dynamic changes of water vapor around the reactor, re-obtain the humidity distribution data and process it to obtain the humidity parameter. In terms of mechanical vibration, accurately acquire the mechanical vibration waveform data of the stirring device and the overall structure of the reactor in the current state, and obtain accurate mechanical vibration parameters through means such as frequency-domain energy analysis. Input the complete set of re-acquired environmental parameters into the pre-trained compensation parameter generation model. This compensation parameter generation model is constructed based on a large amount of previous data and complex algorithms, and can accurately generate an updated set of dynamic compensation parameters according to the new set of environmental parameters. The updated set of dynamic compensation parameters includes temperature compensation coefficients, humidity compensation coefficients, vibration compensation coefficients, etc. required for compensating the original pressure signal for the current environmental change situation.

[0112] Step S330: Perform a moving average fusion on the updated set of dynamic compensation parameters and the historical set of compensation parameters to generate a smoothly transitioning sequence of compensation parameters.

[0113] Step S340: Use the smoothly transitioning sequence of compensation parameters to progressively adjust the originally acquired pressure signals in subsequent acquisitions.

[0114] In a chemical reactor system, the original pressure signal is continuously acquired, reflecting the real-time change of the pressure inside the reactor. When adjusting the original pressure signal using a sequence of compensation parameters with smooth transition, since the sequence of compensation parameters is carefully integrated and updated, it can more accurately take into account the influence of the dynamic changes of environmental parameters on the pressure signal. For example, when the ambient temperature inside the reactor gradually increases, the humidity also fluctuates to a certain extent, and at the same time the mechanical vibration situation changes, the sequence of compensation parameters with smooth transition can make a progressive adjustment to the original pressure signal according to these environmental changes. This progressive adjustment will not cause a sudden jump in the pressure signal, but gradually adjusts the pressure signal according to the gradual change of environmental parameters. For the application of the temperature compensation coefficient, it can gradually correct the pressure signal deviation caused by the temperature increase; the humidity compensation coefficient can smoothly adjust the sensitivity of the pressure signal along with the fluctuation of humidity; the vibration compensation coefficient continuously calibrates the changes in the phase and amplitude of the pressure signal caused by mechanical vibration. Through this progressive adjustment, it can ensure that the pressure signal always maintains a high degree of accuracy under complex and changeable environmental conditions, providing reliable data support for pressure monitoring and control in the chemical production process, and ensuring the safe, stable and efficient operation of chemical production.

[0115] In a possible implementation manner, step S330 includes:

[0116] Step S331, setting the time window length, and arranging the historical compensation parameter set and the updated dynamic compensation parameter set in chronological order within the time window.

[0117] For example, a time window of 30 minutes can be set. Within this time window, the historical compensation parameter set and the updated dynamic compensation parameter set are arranged in chronological order. The historical compensation parameter set is the compensation parameters generated and saved at different time points before, which reflects the parameter situation of compensating the pressure sensor signal under different past environmental conditions. After arranging these two sets of compensation parameters in chronological order within the time window, a weight coefficient related to the timestamp is assigned to each compensation parameter within the time window, where the weight coefficient of the latest parameter is the largest. This is because the latest compensation parameter can better reflect the current environmental state and the actual situation of the pressure sensor. For example, for the newly updated and generated compensation parameter, since it is recalculated based on the current environmental parameters, a larger weight coefficient is given, while for the older compensation parameters within the time window, as time goes by, their weight coefficients gradually decrease. Then, the weighted average of all compensation parameters within the window is calculated, and the calculation process of this weighted average fully considers the weight of each compensation parameter and its corresponding timestamp information. For example, assume that within a 30-minute time window, there are 5 compensation parameters, the weight coefficient of the latest compensation parameter is 0.5, and the weight coefficients of the remaining 4 historical compensation parameters are 0.1, 0.1, 0.2, and 0.1 respectively. After multiplying each compensation parameter by its corresponding weight coefficient and adding them up, the result is the current output value of the smoothly transitioning compensation parameter sequence.

[0118] Step S332: Assign a weight coefficient related to the timestamp to each compensation parameter within the time window, where the weight coefficient of the latest parameter is the largest.

[0119] Step S333: Calculate the weighted average of all compensation parameters within the window as the current output value of the smoothly transitioning compensation parameter sequence.

[0120] Step S334: When a new parameter enters the time window, remove the oldest parameter within the window and recalculate the weighted average to achieve the rolling update of the compensation parameters and generate a smoothly transitioning compensation parameter sequence.

[0121] As the chemical production process continues, new compensation parameters are continuously generated. When a new compensation parameter enters the set time window, the earliest compensation parameter within the time window needs to be removed to keep the number of compensation parameters within the time window unchanged. Then, the weighted average is recalculated according to the previous weight assignment principle. For example, when a new compensation parameter enters, the earliest compensation parameter originally within the window is removed. The weight coefficient of the new compensation parameter is 0.4, and the weight coefficients of the remaining compensation parameters are adjusted according to their relative newness and oldness, such as 0.1, 0.1, 0.2, and 0.2 respectively. The weighted average is calculated again to obtain a new sequence of smoothly transitioning compensation parameters. This rolling update mechanism enables the compensation parameter sequence to adapt to the dynamic changes of environmental parameters in a timely manner and ensures the smooth transition of compensation parameters during the update process, avoiding the problem of inaccurate compensation of pressure signals that may be caused by suddenly replacing compensation parameters.

[0122] In a possible implementation manner, the method further includes:

[0123] Step S410, during the steady-state working stage of the industrial control system, collect the compensated pressure signal output by the target pressure sensor.

[0124] Step S420, synchronously obtain the actual action feedback signal of the actuator.

[0125] In chemical production, the steady-state working stage means that the chemical reaction in the reaction kettle proceeds stably according to the established process parameters. For example, parameters such as temperature, pressure, and material flow rate are in a relatively stable state. At this time, the target pressure sensor continuously monitors the pressure in the reaction kettle and outputs the pressure signal after a series of previous compensation processes. This compensated pressure signal reflects the pressure situation after considering the influence of environmental factors such as temperature, humidity, and mechanical vibration. The actuator, such as the valve motor that controls the material inlet and outlet in the reaction kettle or the heating-cooling device controller that adjusts the temperature of the reaction kettle, will act according to the instructions of the control system, and these actuators themselves have a feedback mechanism and can generate actual action feedback signals. For example, the encoder of the valve motor can feedback the actual opening degree of the valve, and the heating-cooling device controller can feedback the actual heating or cooling power. These actual action feedback signals and the compensated pressure signal are obtained within the same time frame.

[0126] Step S430, perform a correlation analysis on the compensated pressure signal and the actual action feedback signal, and calculate the signal matching degree index.

[0127] Step S440, when the signal matching degree index is lower than the preset safety threshold, trigger the retraining instruction of the compensation parameter generation model and generate a system alarm log.

[0128] In chemical production, the preset safety threshold is set according to the requirements of the production process and past experience, for example, set to 0.8. If the calculated signal matching degree index is lower than 0.8, this may mean that there is a problem with the current compensation parameters, resulting in a deviation in the correlation between the pressure signal and the actual action of the actuator. This deviation may affect the pressure control in the reactor, and further affect the stability of the chemical reaction and the product quality. At this time, the system will automatically trigger a retraining instruction for the compensation parameter generation model. At the same time, to facilitate the operator to understand the abnormal situation of the system, a system alarm log will be generated, recording relevant information such as the time when the signal matching degree index is lower than the safety threshold, the current compensated pressure signal value, and the actual action feedback signal value.

[0129] Step S450, re-collect environmental parameter samples and pressure signal samples according to the retraining instruction, and update the parameter mapping relationship of the compensation parameter generation model.

[0130] After the retraining instruction is triggered, data needs to be re-collected to re-evaluate the relationship between environmental parameters and pressure signals. In the chemical reactor environment, re-collecting environmental parameter samples includes measuring the temperature, humidity, and mechanical vibration around the reactor again accurately. For temperature collection, ensure that the collection points cover all areas around the target pressure sensor that may affect the temperature, and increase the collection frequency to obtain more detailed temperature change conditions. Humidity collection should consider the dynamic changes of water vapor generation in the reactor and workshop ventilation. It may be necessary to increase the number of humidity sensors or use more precise humidity measurement equipment. Mechanical vibration collection requires a comprehensive vibration detection of the stirring device, reactor support structure, etc. to obtain accurate mechanical vibration waveform data. At the same time, the pressure signal samples of the target pressure sensor also need to be re-collected. These pressure signal samples should cover the pressure values under different reaction stages and different environmental conditions. Then, using these newly collected environmental parameter samples and pressure signal samples, according to the training steps of the compensation parameter generation model described above, re-determine the relationship between environmental parameters such as temperature, humidity, and mechanical vibration and the pressure signal error, so as to update the parameter mapping relationship of the compensation parameter generation model, improve the accuracy of the compensation parameters, and ensure a good match between the pressure signal and the actual action of the actuator.

[0131] Among them, step S430 includes:

[0132] Step S431, perform normalization processing on the compensated pressure signal and the actual action feedback signal to eliminate the dimensional difference.

[0133] In a chemical reaction kettle system, the unit of the compensated pressure signal may be Pascal (Pa), while the actual action feedback signal of the actuator may be different dimensions such as the percentage of valve opening or Watt (W) of heating-cooling power. Through normalization, these data with different dimensions are converted into dimensionless relative values, enabling them to be compared on the same mathematical scale. For example, for the pressure signal, its value can be mapped to between 0 and 1, and for the valve opening feedback signal, it can also be mapped to between 0 and 1.

[0134] Step S432: Calculate the cross-correlation coefficient matrix of the compensated pressure signal and the actual action feedback signal within the same time window, and determine the maximum correlation coefficient and its corresponding time delay.

[0135] In the chemical production process, due to factors such as signal transmission delay and the response time of the actuator, there may be a certain time difference between the compensated pressure signal and the actual action feedback signal. For example, when the pressure in the reaction kettle increases, the control system issues an instruction to open the valve to reduce the pressure based on the compensated pressure signal, but there will be a short delay from when the valve receives the instruction to when it actually starts to open, and this delay will be reflected in the time delay of the signal. By calculating the cross-correlation coefficient matrix, the degree of correlation between the two signals under different time delays can be found. For example, assuming the time window is set to 10 minutes and the cross-correlation coefficient is calculated at 1-second intervals, a cross-correlation coefficient matrix corresponding to different time delays within 10 minutes can be obtained. In this cross-correlation coefficient matrix, there will be a maximum correlation coefficient, and the corresponding time delay is the time difference of the best match between the two signals.

[0136] Step S433: Based on the maximum correlation coefficient and the time delay information, construct a signal consistency evaluation function, and use the output value of the signal consistency evaluation function as the signal matching degree index.

[0137] The signal consistency evaluation function is constructed based on the characteristics of the chemical production process and the requirements for signal matching. For example, it can be a function that considers the weights of the maximum correlation coefficient and the time delay, such as signal matching degree index = maximum correlation coefficient × (1 - time delay weight × absolute value of time delay). Among them, the time delay weight is set according to the requirements of the chemical production for signal real-time performance. If the requirement for real-time performance is very high, the time delay weight will be larger. This signal matching degree index comprehensively reflects the matching degree between the compensated pressure signal and the actual action feedback signal, and its value is between 0 and 1. The closer it is to 1, the higher the matching degree of the two signals.

[0138] In a possible implementation manner, the method further includes:

[0139] Step S510: Real-time monitor the signal output stability index of the target pressure sensor.

[0140] In the complex environment of a chemical production workshop, the signal output stability of the target pressure sensor is affected by various factors. For example, the electronic components of the sensor itself may age or malfunction due to long-term operation, the surrounding electromagnetic environment interference may affect the signal transmission stability, and environmental factors such as temperature, humidity, and mechanical vibration may also cause fluctuations in the sensor's performance. By real-time monitoring the signal output stability index, these potential problems can be detected in a timely manner. This signal output stability index can be a comprehensive index constructed based on factors such as the amplitude fluctuation range and frequency stability of the signal.

[0141] Step S520, when it is detected based on the signal output stability index that the signal output is interrupted or the amplitude exceeds the physical range, start the data acquisition channel of the backup pressure sensor.

[0142] In a chemical reactor system, if the signal output of the target pressure sensor is interrupted, this may be due to a wiring fault, power supply problem, or a serious internal fault of the sensor. And the amplitude exceeding the physical range may be because of an abnormal high pressure in the reactor or a measurement deviation of the sensor itself. Once such a situation is detected, to ensure the continuity of pressure monitoring, the system will immediately start the data acquisition channel of the backup pressure sensor. The backup pressure sensor exists as a redundant device, and its installation location and measurement principle are similar to those of the target pressure sensor and it has been pre-calibrated.

[0143] Step S530, connect the output signal of the backup pressure sensor to the signal processing module of the industrial control system after being processed by the same compensation process.

[0144] The output signal of the backup pressure sensor is also affected by environmental factors such as temperature, humidity, and mechanical vibration, so it needs to be processed by the same compensation process as the target pressure sensor. For example, first obtain the original pressure signal sequence of the backup pressure sensor within a preset acquisition period, then detect the set of real-time environmental parameters around it, including temperature, humidity, and mechanical vibration parameters, input these environmental parameters into a pre-trained compensation parameter generation model to generate a set of dynamic compensation parameters, and then adjust the original pressure signal step by step according to this set of dynamic compensation parameters to obtain the compensated pressure signal. Finally, connect the compensated pressure signal of the backup pressure sensor to the signal processing module of the industrial control system to replace the signal of the target pressure sensor and ensure the normal operation of the pressure monitoring and control system.

[0145] Step S540, synchronously record the abnormal state information of the target pressure sensor and generate a device maintenance request instruction.

[0146] After detecting an abnormality in the target pressure sensor, it is necessary to record in detail the relevant abnormal status information, such as the time of signal interruption, the specific value of the amplitude exceeding the range, and the signal fluctuation conditions during a previous period of time. This information helps maintenance personnel quickly locate problems. At the same time, a device maintenance request instruction is generated to notify the relevant maintenance personnel to repair the target pressure sensor.

[0147] Step S550, after the device maintenance is completed, perform a full-scale calibration test on the target pressure sensor, and re-enable it after verifying the effectiveness of the compensation parameter generation model.

[0148] For example, device maintenance may involve operations such as repairing the sensor, replacing parts, or recalibrating. After the maintenance is completed, a full-scale calibration test needs to be performed on the target pressure sensor. In the environment of a chemical reactor, the full-scale calibration test requires the use of a high-precision pressure source. Within the entire measurement range of the target pressure sensor, known standard pressure values are gradually applied from the minimum value to the maximum value. At the same time, the output signal of the sensor is recorded and compared with the standard pressure value to calculate the error. If the error is within an acceptable range, it indicates that the sensor has returned to the normal working state. Then, the target pressure sensor is reconnected to the system, and its output pressure signal is collected again. The compensation process is carried out according to the normal procedure, and the correlation between the compensated pressure signal and the actual action feedback signal of the actuator is analyzed to verify the effectiveness of the compensation parameter generation model. If the verification passes, it means that the compensation parameter generation model can accurately compensate the signal of the target pressure sensor. At this time, the target pressure sensor can be officially re-enabled for normal pressure monitoring work.

[0149] Figure 2 FIG. shows a schematic diagram of exemplary hardware and software components of a pressure sensor data accuracy compensation system 100 for industrial control that can implement the ideas of the present application provided by some embodiments of the present application. For example, the processor 120 can be used on the pressure sensor data accuracy compensation system 100 for industrial control and is used to execute the functions in the present application.

[0150] The pressure sensor data accuracy compensation system 100 for industrial control can be a general-purpose server or a special-purpose server, both of which can be used to implement the pressure sensor data accuracy compensation method for industrial control of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0151] For example, a pressure sensor data accuracy compensation system 100 for industrial control may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the pressure sensor data accuracy compensation system 100 for industrial control may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The pressure sensor data accuracy compensation system 100 for industrial control further includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0152] For ease of explanation, only one processor is described in the pressure sensor data accuracy compensation system 100 for industrial control. However, it should be noted that the pressure sensor data accuracy compensation system 100 for industrial control in the present application may also include multiple processors. Therefore, the steps performed by one processor described in the present application may also be jointly performed or separately performed by multiple processors. For example, if the processor of the pressure sensor data accuracy compensation system 100 for industrial control performs step A and step B, it should be understood that step A and step B may also be jointly performed by two different processors or separately performed in one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0153] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned pressure sensor data accuracy compensation method for industrial control is implemented.

[0154] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.

Claims

1. A pressure sensor data accuracy compensation method for industrial control, characterized in that: The method comprises: Acquire an original pressure signal sequence output by a target pressure sensor within a preset acquisition period, wherein the original pressure signal sequence includes a plurality of pressure signals arranged by timestamps; Detecting a set of real-time environmental parameters of the target pressure sensor within the acquisition period, wherein the set of real-time environmental parameters includes a temperature parameter, a humidity parameter, and a mechanical vibration parameter; Inputting the real-time environmental parameter set into a pre-trained compensation parameter generation model to generate a dynamic compensation parameter set corresponding to the original pressure signal sequence; According to the dynamic compensation parameter set, each pressure signal in the original pressure signal sequence is adjusted step by step to obtain a compensated pressure signal sequence; Outputting the compensated pressure signal sequence to a signal processing module of an industrial control system for driving an actuator of the industrial control system; The step of adjusting each pressure signal in the original pressure signal sequence step by step according to the dynamic compensation parameter set to obtain a compensated pressure signal sequence includes: Performing time domain filtering on the original pressure signal sequence to obtain a preprocessed pressure signal sequence; performing baseline offset correction on the preprocessed pressure signal sequence according to the temperature compensation coefficient in the dynamic compensation parameter set; Performing sensitivity scaling adjustment on the pressure signal sequence after the baseline offset correction according to the humidity compensation coefficient in the dynamic compensation parameter set; performing phase synchronization calibration on the pressure signal sequence after the sensitivity scaling adjustment according to the vibration compensation coefficient in the dynamic compensation parameter set; The signal sequence after phase synchronization calibration is used as the compensated pressure signal sequence.

2. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The detecting a set of real-time environmental parameters of the target pressure sensor within the acquisition period includes: The temperature distribution data in the space where the target pressure sensor is located is collected by a temperature sensor array, and a spatial weighted average processing is performed on the temperature distribution data to obtain the temperature parameter; Collecting humidity distribution data in the space where the target pressure sensor is located through a humidity sensor array, and performing time series filtering on the humidity distribution data to obtain the humidity parameter; The mechanical vibration waveform data of the target pressure sensor installation position is collected by a vibration sensor, and the frequency domain energy analysis is performed on the mechanical vibration waveform data to extract the energy peak value within a preset frequency band as the mechanical vibration parameter; The data acquisition timestamps of the temperature parameter, humidity parameter and mechanical vibration parameter are synchronized and aligned with the timestamp of the original pressure signal sequence.

3. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The step of inputting the real-time environmental parameter set into a pre-trained compensation parameter generation model to generate a dynamic compensation parameter set corresponding to the original pressure signal sequence includes: Performing timestamp alignment verification on the temperature parameters, humidity parameters and mechanical vibration parameters in the real-time environmental parameter set, eliminating environmental parameter data whose timestamp deviation exceeds a preset synchronization threshold, and generating a temperature parameter sequence, a humidity parameter sequence and a mechanical vibration parameter sequence after synchronization calibration; Input the synchronously calibrated temperature parameter sequence into the temperature feature extraction channel of the compensation parameter generation model, and extract the temperature impact feature vector through the three-layer convolution network of the temperature feature extraction channel, wherein the first layer of convolution network captures the low-frequency trend component of the temperature parameter, the second layer of convolution network extracts the medium-frequency disturbance component of the temperature fluctuation, and the third layer of convolution network identifies the transient impact component of the temperature mutation; Inputting the synchronously calibrated humidity parameter sequence into the humidity feature extraction channel of the compensation parameter generation model, generating a humidity impact feature vector through the bidirectional recurrent network of the humidity feature extraction channel, wherein the forward propagation process captures the cumulative penetration effect of humidity parameters on the sensor material, and the reverse propagation process identifies the hysteresis compensation demand caused by humidity mutation; The synchronously calibrated mechanical vibration parameter sequence is input into the vibration feature extraction channel of the compensation parameter generation model, and the vibration influence feature vector is decomposed through the time-frequency hybrid network of the vibration feature extraction channel, wherein the time domain analysis unit extracts the envelope characteristics of the vibration amplitude, and the frequency domain analysis unit separates the distribution ratio of the vibration energy at different resonant frequency points; The temperature influence feature vector, humidity influence feature vector and vibration influence feature vector are integrated, and the coupling strength matrix of the three is calculated through the feature cross layer of the compensation parameter generation model, wherein the temperature-humidity coupling strength is characterized by the dot product value of the two-way feature vector, the humidity-vibration coupling strength is calculated by the covariance in the sliding window, and the temperature-vibration coupling strength is evaluated by the maximum mutual information entropy; Dynamically allocating compensation weight coefficients according to the coupling strength matrix, generating a real-time proportional combination of temperature compensation weight, humidity compensation weight and vibration compensation weight through the weight allocator of the compensation parameter generation model, wherein the temperature compensation weight is positively correlated with the temperature-humidity coupling strength, the humidity compensation weight is exponentially related to the humidity-vibration coupling strength, and the vibration compensation weight forms a dynamic feedback adjustment with the temperature-vibration coupling strength; The temperature compensation weight, humidity compensation weight and vibration compensation weight are input into the multi-layer compensation parameter generator of the compensation parameter generation model, a basic offset of the temperature compensation coefficient is generated through the first compensation layer, a sensitivity adjustment amount of the humidity compensation coefficient is superimposed on the second compensation layer, and a phase correction amount of the vibration compensation coefficient is injected into the third compensation layer, and finally a dynamic compensation parameter set including the temperature compensation coefficient, the humidity compensation coefficient and the vibration compensation coefficient is output; Verify the physical feasibility boundary of the dynamic compensation parameter set, detect whether each compensation coefficient exceeds the safety threshold of the sensor range through the parameter constraint module of the compensation parameter generation model, truncate the compensation coefficient that exceeds the safety threshold and generate a compensation parameter correction flag, and match and align the corrected dynamic compensation parameter set with the timestamp of the original pressure signal sequence.

4. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The training step of the compensation parameter generation model includes: Under the conditions of multiple different environmental parameter combinations, collecting a pressure signal sample set of the target pressure sensor and a corresponding standard pressure signal set; For each combination of environmental parameters, calculating an error distribution matrix between the pressure signal sample set and the standard pressure signal set; Performing nonlinear regression fitting on the environmental parameter combination and the error distribution matrix to determine a parameter mapping relationship in the compensation parameter generation model; Iteratively optimizing the weight coefficients in the parameter mapping relationship by a back propagation algorithm until the matching degree between the predicted compensation parameters output by the compensation parameter generation model and the error distribution matrix reaches a preset threshold; The step of performing nonlinear regression fitting on the environmental parameter combination and the error distribution matrix to determine the parameter mapping relationship in the compensation parameter generation model includes: Performing three-dimensional surface fitting on the temperature parameter, humidity parameter and mechanical vibration parameter in the environmental parameter combination and the error amplitude of the corresponding time stamp in the error distribution matrix, respectively, to generate an initial parameter mapping relationship including a temperature compensation surface, a humidity compensation surface and a vibration compensation surface; Calculating the error change rate of each pressure signal sample in a continuous acquisition cycle to generate a dynamic gradient vector of the error distribution matrix; Performing a multivariate coupling analysis on the temperature parameter change rate, humidity parameter change rate and mechanical vibration parameter change rate in the environmental parameter combination and the dynamic gradient vector to generate a dynamic compensation coefficient set including a temperature gradient weight, a humidity gradient weight and a vibration gradient weight; Performing point-by-point multiplication operations on the temperature compensation surface, the humidity compensation surface and the vibration compensation surface in the initial parameter mapping relationship and the corresponding gradient weights in the dynamic compensation coefficient set, respectively, to generate an enhanced parameter mapping relationship integrating static error features and dynamic error features; A neural network model with a multi-layer hidden structure is constructed, and the temperature compensation surface data, humidity compensation surface data, and vibration compensation surface data in the enhanced parameter mapping relationship are respectively input into three independent input channels of the neural network model, and nonlinear feature fusion is performed on the data of the three input channels through the fully connected layer of the neural network model to generate a final parameter mapping relationship with a cross-compensation effect; Optimizing the output accuracy of the final parameter mapping relationship by performing residual back propagation calculation on the output layer data of the neural network model and the standard pressure signal set, and adjusting the node connection weights of the fully connected layer in the neural network model until the prediction error value of the final parameter mapping relationship reaches a preset accuracy threshold; The optimized node connection weights in the neural network model are solidified to generate a data structure of executable parameter mapping relationships in the compensation parameter generation model, wherein the data structure includes a cross-query interface for a temperature compensation surface index table, a humidity compensation surface index table, and a vibration compensation surface index table.

5. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The step of performing phase synchronization calibration on the pressure signal sequence after the sensitivity scaling adjustment according to the vibration compensation coefficient in the dynamic compensation parameter set includes: Extracting the vibration frequency component corresponding to the mechanical vibration parameter, and generating a reference sinusoidal waveform having the same period as the vibration frequency component; Performing local extreme point detection on the pressure signal sequence after the sensitivity scaling adjustment to determine the phase offset of the signal waveform; Performing phase compensation on the reference sinusoidal waveform according to the phase offset and the vibration compensation coefficient in the dynamic compensation parameter set to generate an anti-vibration interference template; The anti-vibration interference template is convolved with the pressure signal sequence after the sensitivity scaling adjustment to suppress the signal distortion component caused by vibration.

6. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The method further comprises: Monitoring the change gradient of the environmental parameters of the target pressure sensor during a continuous acquisition cycle; When the environmental parameter change gradient exceeds a preset fluctuation threshold, the current environmental parameter set is re-collected and input into the compensation parameter generation model to generate an updated dynamic compensation parameter set; Perform sliding average fusion of the updated dynamic compensation parameter set and the historical compensation parameter set to generate a compensation parameter sequence with a smooth transition; The smooth-transition compensation parameter sequence is used to progressively adjust the subsequently collected original pressure signal.

7. The pressure sensor data accuracy compensation method for industrial control according to claim 6, characterized in that: The step of performing sliding average fusion on the updated dynamic compensation parameter set and the historical compensation parameter set to generate a compensation parameter sequence with a smooth transition includes: Setting a time window length, and arranging the historical compensation parameter set and the updated dynamic compensation parameter set in chronological order within the time window; Assigning a timestamp-related weight coefficient to each compensation parameter within the time window, wherein the weight coefficient of the latest parameter is the largest; Calculating a weighted average value of all compensation parameters in the window as the current output value of the compensation parameter sequence for the smooth transition; When a new parameter enters the time window, the oldest parameter in the window is removed and the weighted average is recalculated to achieve a rolling update of the compensation parameter and generate a smooth transition compensation parameter sequence.

8. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The method further comprises: In a steady-state operation phase of the industrial control system, collecting a compensated pressure signal output by the target pressure sensor; Synchronously obtaining an actual action feedback signal of the actuator; Performing correlation analysis on the compensated pressure signal and the actual action feedback signal to calculate a signal matching index; When the signal matching index is lower than a preset safety threshold, a retraining instruction of the compensation parameter generation model is triggered, and a system alarm log is generated; Recollect environmental parameter samples and pressure signal samples according to the retraining instruction, and update the parameter mapping relationship of the compensation parameter generation model; The step of performing correlation analysis on the compensated pressure signal and the actual action feedback signal to calculate a signal matching index includes: Normalizing the compensated pressure signal and the actual action feedback signal to eliminate dimensional differences; Calculating the mutual correlation coefficient matrix of the compensated pressure signal and the actual action feedback signal in the same time window, and determining the maximum correlation coefficient and its corresponding delay; A signal consistency evaluation function is constructed according to the maximum correlation coefficient and the time delay information, and an output value of the signal consistency evaluation function is used as the signal matching index.

9. The pressure sensor data accuracy compensation method for industrial control according to claim 1, characterized in that: The method further comprises: Real-time monitoring of the signal output stability index of the target pressure sensor; When it is detected based on the signal output stability indicator that the signal output is interrupted or the amplitude exceeds the physical range, the data acquisition channel of the standby pressure sensor is started; The output signal of the backup pressure sensor is processed by the same compensation process and then connected to the signal processing module of the industrial control system; synchronously recording abnormal status information of the target pressure sensor and generating an equipment maintenance request instruction; After the equipment maintenance is completed, the target pressure sensor is subjected to a full-scale calibration test to verify the validity of the compensation parameter generation model and then reactivated.

10. A pressure sensor data accuracy compensation system for industrial control, characterized in that: The pressure sensor data accuracy compensation system for industrial control includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the pressure sensor data accuracy compensation method for industrial control described in any one of claims 1 to 9.

Images