Robot explosion-proof dynamic threshold regulating method and system

Abstract

The application relates to a robot explosion-proof dynamic threshold regulation method and system, which comprises the following steps: collecting pressure data in a robot operation environment in real time through a high-precision pressure sensor; inputting the pressure data into a long short-term memory network model which has been pre-trained to perform time series prediction processing, so as to generate a pressure prediction value in a future preset time period; calculating a dynamic explosion-proof threshold according to the pressure prediction value, wherein the threshold changes with time and is associated with the pressure prediction value; determining a current risk level according to the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold, and then adjusting an explosion-proof control strategy based on the risk level, generating corresponding explosion-proof control instructions and sending the instructions to the robot to realize dynamic regulation; the scheme improves the safety and adaptability of the robot in an explosion-proof environment, effectively reduces the safety hidden danger caused by the change of the environmental pressure, and improves the production efficiency on the premise of ensuring safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control and robotics, and particularly relates to a robot explosion-proof dynamic threshold regulation method and system. BACKGROUND

[0002] In industrial production, the application scenarios of robots are increasingly widespread, especially in some environments with flammable and explosive risks, such as petroleum and chemical industry, coal mining and other industries. The explosion-proof safety of robots is crucial. The traditional robot explosion-proof technology mainly relies on fixed explosion-proof threshold settings. This method ensures the safe operation of robots to a certain extent.

[0003] The setting method of the fixed explosion-proof threshold is based on experience and general safety standards, and a fixed pressure value is set as the explosion-proof limit in advance. When the pressure in the robot operating environment exceeds this fixed value, the corresponding explosion-proof measures are triggered. However, this method has obvious limitations, as it cannot adapt to the dynamic changes of environmental pressure. In actual production processes, environmental pressure is affected by various factors, such as changes in process flow, running state of equipment, etc., resulting in continuous fluctuations in pressure values.

[0004] Using a fixed explosion-proof threshold cannot be adjusted according to the real-time changes of environmental pressure, which may result in two situations. First, when the environmental pressure fluctuates greatly, the fixed threshold may not accurately reflect the actual explosion-proof needs in time, leading to excessive or insufficient explosion-proof measures, affecting production efficiency and safety. Second, for some complex pressure changes in the environment, the fixed threshold is difficult to cover all possible dangerous situations, which may pose a certain safety risk. SUMMARY

[0005] The main purpose of the present application is to provide a robot explosion-proof dynamic threshold regulation method and system, which can be adjusted according to the real-time changes of environmental pressure, improving the safety and adaptability of robots in explosion-proof environments.

[0006] To achieve the above purpose, the robot explosion-proof dynamic threshold regulation method provided by the embodiments of the present application comprises:

[0007] Real-time acquisition of pressure data in the robot operating environment by a high-precision pressure sensor, wherein the pressure data includes pressure values at multiple time points;

[0008] Inputting the pressure data into a pre-trained long short-term memory network model for time series prediction processing to generate pressure prediction values in a future preset time period;

[0009] Calculating a dynamic explosion-proof threshold according to the pressure prediction values, wherein the dynamic explosion-proof threshold changes with time and is associated with the pressure prediction values;

[0010] determine a current risk level according to a relationship between an actual value of the pressure data and the dynamic explosion-proof threshold value;

[0011] adjust an explosion-proof control strategy based on the current risk level, generate a corresponding explosion-proof control instruction, and send the explosion-proof control instruction to the robot to realize dynamic regulation and control.

[0012] Correspondingly, the embodiment of the application also provides a robot explosion-proof dynamic threshold regulation and control system, the system comprising:

[0013] An acquisition module is configured to collect pressure data in a running environment of a robot in real time through a high-precision pressure sensor, wherein the pressure data comprises pressure values at multiple time points.

[0014] A prediction processing module is configured to input the pressure data into a long short-term memory network model that has been pre-trained to perform time series prediction processing, and generate pressure prediction values in a future preset time period.

[0015] A threshold calculation module is configured to calculate a dynamic explosion-proof threshold value according to the pressure prediction values, wherein the dynamic explosion-proof threshold value changes over time and is associated with the pressure prediction values.

[0016] A risk determination module is configured to determine a current risk level according to a relationship between an actual value of the pressure data and the dynamic explosion-proof threshold value.

[0017] A control module is configured to adjust an explosion-proof control strategy based on the current risk level, generate a corresponding explosion-proof control instruction, and send the explosion-proof control instruction to the robot to realize dynamic regulation and control.

[0018] In summary, by using the technical solution of the application, the pressure data in the running environment of the robot can be collected in real time through the high-precision pressure sensor, and the dynamic change of the pressure can be accurately obtained. The pressure data is input into the long short-term memory network model that has been pre-trained to perform time series prediction processing, and the pressure prediction values in the future preset time period are generated, so that the trend of the pressure change can be predicted in advance. The dynamic explosion-proof threshold value is calculated according to the pressure prediction values, the threshold value changes over time and is associated with the pressure prediction values, which can more accurately adapt to the dynamic change of the environmental pressure and avoid the problem that the fixed threshold value cannot accurately reflect the actual explosion-proof demand in time. The current risk level is determined according to the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold value, and then the explosion-proof control strategy is adjusted based on the risk level, a corresponding explosion-proof control instruction is generated, and the explosion-proof control instruction is sent to the robot to realize dynamic regulation and control, which improves the safety and adaptability of the robot in the explosion-proof environment, effectively reduces the safety hazards caused by the change of the environmental pressure, and also improves the production efficiency under the premise of ensuring safety. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0020] Figure 1 is a scene diagram of the robot explosion-proof dynamic threshold regulation method in the embodiments of the present application;

[0021] Figure 2 is a flowchart of the robot explosion-proof dynamic threshold regulation method provided by the embodiments of the present application;

[0022] Figure 3 is a flowchart of the explosion-proof threshold generation provided by the embodiments of the present application;

[0023] Figure 4 is a flowchart of the correction coefficient generation provided by the embodiments of the present application;

[0024] Figure 5 is a flowchart of the explosion-proof threshold adjustment provided by the embodiments of the present application;

[0025] Figure 6 is a flowchart of the risk determination provided by the embodiments of the present application;

[0026] Figure 7 is a flowchart of the explosion-proof control provided by the embodiments of the present application;

[0027] Figure 8 is a flowchart of the dependency relationship extraction provided by the embodiments of the present application

[0028] Figure 9 is a flowchart of the hidden layer processing provided by the embodiments of the present application;

[0029] Figure 10 is a structural diagram of the robot explosion-proof dynamic threshold regulation system provided by the embodiments of the present application;

[0030] Figure 11 is a structural diagram of the computer device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0032] The embodiments of the present application provide a robot explosion-proof dynamic threshold regulation method and system, which will be described in detail below.

[0033] In the application embodiments, the robot explosion-proof dynamic threshold regulation based on LSTM time series prediction is an intelligent robot explosion-proof control method. It uses a long short-term memory (LSTM) model to perform time series prediction on pressure data in the robot operating environment, calculates a dynamically changing explosion-proof threshold according to the prediction result to adapt to the real-time changes of the environmental pressure. Among them, a high-precision pressure sensor is used to collect pressure data in the robot operating environment in real time, and these data contain pressure values at multiple time points. The long short-term memory model can capture the long-term dependence of pressure changes through learning of historical pressure data, thereby accurately predicting the pressure values in the future preset time period. The dynamic explosion-proof threshold is calculated according to the pressure prediction value, which will be dynamically adjusted with the passage of time and the change of the pressure prediction value. By comparing the actual value of the pressure data with the dynamic explosion-proof threshold, the current risk level is determined, and the explosion-proof control strategy is adjusted according to the risk level, and the corresponding explosion-proof control instruction is generated and sent to the robot, to realize the dynamic regulation of the robot explosion-proof, and improve the safety and adaptability of the robot in the flammable and explosive environment.

[0034] As shown in Figure 1 , a robot explosion-proof dynamic threshold regulation method based on LSTM time series prediction is provided, and in this scene, it mainly contains a robot, a high-precision pressure sensor, and a control platform. The robot, the high-precision pressure sensor, and the control platform are connected through a wireless network.

[0035] Taking the petroleum and chemical production workshop scene as an example, there are a large number of flammable and explosive chemicals in this scene, and the robot undertakes important tasks such as equipment inspection and material handling. In the production process, the pressure in the workshop will fluctuate with changes in the process flow, the running state of the equipment and other factors. If the explosion-proof control strategy of the robot cannot adapt to such pressure changes in time, it may lead to safety accidents.

[0036] The high-precision pressure sensor is installed at key positions around the robot and can collect pressure data in the robot operating environment in real time. For example, when the robot is inspecting the equipment, the pressure sensor will continuously record the pressure values around the equipment; during the material handling process, it will also record the pressure changes in the handling area. These pressure data contain pressure values at multiple time points and can reflect the dynamic changes of the pressure.

[0037] The control platform is responsible for receiving and processing pressure data uploaded by high-precision pressure sensors. First, the control platform inputs the pressure data into a pre-trained long short-term memory network model for time series prediction processing, generating pressure prediction values for a future preset time period. For example, by learning from pressure data over a period of time, the model can predict the pressure trend in the next 10 minutes, 30 minutes, or 1 hour.

[0038] Then, the control platform calculates the dynamic explosion-proof threshold based on the pressure prediction values. This process includes extracting the maximum and minimum predicted pressure values from the pressure prediction values, calculating the pressure fluctuation amplitude, calculating the initial explosion-proof threshold based on the pressure fluctuation amplitude and the preset safety margin coefficient, and correcting the initial explosion-proof threshold based on the statistical characteristics of historical environmental pressure data to obtain the final dynamic explosion-proof threshold.

[0039] Next, the control platform determines the current risk level based on the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold. If the difference between the actual value of the pressure data and the dynamic explosion-proof threshold is in the low-risk interval, the current risk level is determined to be level one; if the difference is in the medium-risk interval, the current risk level is determined to be level two; if the difference is in the high-risk interval, the current risk level is determined to be level three.

[0040] Finally, the control platform adjusts the explosion-proof control strategy based on the current risk level, generates corresponding explosion-proof control instructions, and sends them to the robot. The robot adjusts the working state of the pressure regulation device according to the explosion-proof control instructions, such as adjusting the pressure release rate, pressure regulation period, and pressure upper limit value, to ensure safe operation in different pressure environments. For example, when the risk level is one, the robot can appropriately reduce the working intensity of the pressure regulation device to improve production efficiency; when the risk level is three, the robot needs to take immediate emergency explosion-proof measures to ensure the safety of itself and the surrounding environment.

[0041] Reference Figure 2 , Figure 2 is a flowchart of a robot explosion-proof dynamic threshold regulation method provided by the present application. The execution subject of the method can be a computer device (which can serve as a control platform). The computer device can be a single computer device or a cluster of multiple computer devices. The computer device can be a terminal device or a server, etc. The robot explosion-proof dynamic threshold regulation method provided by the present application specifically includes:

[0042] S10: Real-time acquisition of pressure data in the robot operating environment by a high-precision pressure sensor, wherein the pressure data includes pressure values at multiple time points.

[0043] In the embodiments of the present application, the high-precision pressure sensor is a device capable of accurately measuring pressure. It can convert pressure signals into electrical signals and record and transmit them through a data acquisition system. Pressure data refers to the pressure values corresponding to different time points in the robot operating environment. These data can reflect the changes in pressure over time.

[0044] In flammable and explosive environments where robots operate, such as oil and chemical plants, coal mine tunnels, etc., environmental pressure will change due to various factors. For example, in an oil and chemical plant, the start and stop of equipment, the progress of chemical reactions, etc. can cause fluctuations in pressure. High-precision pressure sensors can monitor these pressure changes in real time, providing accurate data for subsequent analysis and processing.

[0045] In an embodiment, a high-precision pressure sensor based on the piezoresistive effect can be selected. This sensor utilizes the piezoresistive effect of semiconductor materials, which changes the resistance of the semiconductor material when subjected to pressure. By measuring the change in resistance, the size of the pressure can be determined. The sensor is installed at key locations around the robot, such as near flammable and explosive equipment or pipelines, to ensure that accurate environmental pressure data can be collected. The sensor transmits the collected pressure data to the control platform through wired or wireless means, and the control platform stores and processes the data. By collecting pressure data in real time, abnormal changes in pressure can be detected in a timely manner, providing an important basis for ensuring the safe operation of the robot.

[0046] S20: input the pressure data into the pre-trained long short-term memory network model for time series prediction processing to generate pressure prediction values in a future preset time period.

[0047] In the embodiments of the present application, the long short-term memory network (LSTM) model is a special recurrent neural network that can effectively handle long-term dependencies in sequence data. Time series prediction processing refers to the process of predicting data in a future period of time using historical data.

[0048] In the embodiments of the present application, the pressure values in the future preset time period refer to the pressure values that may occur in a specific future time period obtained by time series prediction of pressure data in the robot operating environment using the long short-term memory network (LSTM) model.

[0049] The preset time period is a time range set in advance according to actual needs, such as 10 minutes, 30 minutes, or 1 hour in the future, etc. The pressure value is not the current pressure value actually measured, but the result of predicting the pressure change trend in the future period of time based on historical pressure data and the pressure change law and long-term dependence learned by the LSTM model. These predicted pressure values can help to understand the possible changes in the robot operating environment in advance, provide an important basis for calculating the dynamic explosion-proof threshold, evaluating the risk level, and formulating the corresponding explosion-proof control strategy, thereby improving the safety and adaptability of the robot in the flammable and explosive environment. In the robot operating environment, pressure changes usually have certain regularity and long-term dependence. For example, certain production processes may cause the pressure to change periodically, or the pressure may show an upward or downward trend over a period of time. The LSTM model can capture these long-term dependencies by learning from historical pressure data, thereby accurately predicting the pressure change in the future preset time period.

[0050] In an embodiment, the collected pressure data is first preprocessed, including data cleaning, normalization, etc. Then the preprocessed data is input into the LSTM model trained in advance. The LSTM model consists of an input layer, a hidden layer, and an output layer. The input layer receives the pressure data, the hidden layer learns the features of the data through LSTM units, extracts the long-term dependencies in the time series, and the output layer generates the pressure prediction value in the future preset time period. Finally, the prediction value is de-normalized to obtain the prediction result within the actual pressure value range. By using the LSTM model for time series prediction processing, the change trend of the pressure can be understood in advance, providing strong support for subsequent explosion-proof control.

[0051] S30: calculating a dynamic explosion-proof threshold according to the pressure prediction value, the dynamic explosion-proof threshold varying with time and being associated with the pressure prediction value.

[0052] In the embodiment of the present application, the dynamic explosion-proof threshold refers to the explosion-proof safety limit calculated according to the pressure prediction value and varying with time. It is associated with the pressure prediction value and can be dynamically adjusted according to the change trend of the pressure to adapt to different environmental pressure conditions.

[0053] During the operation of the robot, the environmental pressure is constantly changing, and a fixed explosion-proof threshold cannot meet the actual explosion-proof needs. The calculation of the dynamic explosion-proof threshold takes into account the prediction value of the pressure, which can more accurately reflect the explosion-proof requirements of the current environment. For example, when it is predicted that the pressure will rise, the dynamic explosion-proof threshold will be correspondingly increased to ensure safety in the case of rising pressure; when it is predicted that the pressure will drop, the dynamic explosion-proof threshold will be appropriately reduced to avoid excessive explosion-proof measures affecting production efficiency.

[0054] In an embodiment, the process of calculating the dynamic explosion-proof threshold is as follows: first, the maximum predicted pressure value and the minimum predicted pressure value are extracted from the pressure prediction value, and the difference between them is calculated as the pressure fluctuation amplitude. Then, based on the pressure fluctuation amplitude and the preset safety margin coefficient, the initial explosion-proof threshold of the dynamic explosion-proof threshold is calculated. Finally, the initial explosion-proof threshold is corrected according to the statistical characteristics of the historical environmental pressure data to obtain the final dynamic explosion-proof threshold. In this way, the dynamic explosion-proof threshold calculated in this way can better adapt to the changes of the environmental pressure and improve the explosion-proof safety of the robot.

[0055] S40: determining the current risk level according to the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold.

[0056] In the embodiments of the present application, the risk level is divided according to the difference between the actual value of the pressure data and the dynamic explosion-proof threshold, which is used to evaluate the explosion-proof risk degree of the current environment.

[0057] In the robot operating environment, by comparing the actual value of the pressure data with the dynamic explosion-proof threshold, it can be judged whether the current pressure state is within the safe range. If the actual value is close to or exceeds the dynamic explosion-proof threshold, it means that there is a high explosion-proof risk; if the actual value is much lower than the dynamic explosion-proof threshold, it means that the risk is low. According to different risk levels, risk levels are divided, which helps to take corresponding explosion-proof control measures.

[0058] In an embodiment, first, the difference between the actual value of the pressure data and the dynamic explosion-proof threshold is obtained, which is used to represent the degree of deviation of the current pressure state from the safe range. Then, the difference is mapped to a preset risk level interval, which includes a low risk interval, a medium risk interval and a high risk interval. If the difference is in the low risk interval, the current risk level is determined to be level one; if the difference is in the medium risk interval, it is determined to be level two; if the difference is in the high risk interval, it is determined to be level three. By determining the risk level, the current explosion-proof risk situation can be understood in time, which provides a basis for subsequent explosion-proof control.

[0059] S50: adjusting the explosion-proof control strategy based on the current risk level, generating corresponding explosion-proof control instructions and sending them to the robot to realize dynamic regulation and control.

[0060] In the embodiments of the present application, the explosion-proof control strategy refers to a series of control measures taken to ensure the safe operation of the robot in the flammable and explosive environment. The explosion-proof control instruction is a specific operation instruction generated according to the explosion-proof control strategy, which is used to control the pressure regulating device and other equipment of the robot.

[0061] Different explosion-proof control measures are required at different risk levels. For example, at low risk levels, the intensity of explosion-proof control can be appropriately reduced to improve production efficiency; at high risk levels, immediate emergency explosion-proof measures are necessary to ensure the safety of the robot and its surrounding environment. By adjusting the explosion-proof control strategy based on the risk level, dynamic control of robot explosion-proof measures can be achieved.

[0062] In one embodiment, corresponding explosion-proof control parameters, such as pressure release rate, pressure regulation cycle, and pressure upper limit, are extracted from a preset explosion-proof strategy library based on the current risk level. Then, explosion-proof control instructions are generated based on these parameters, including specific execution actions and execution times. Finally, the instructions are sent to the robot, which adjusts the operating state of the pressure regulating device accordingly. In this way, the explosion-proof control strategy can be dynamically adjusted according to the real-time risk level, improving the robot's explosion-proof safety and adaptability.

[0063] In one embodiment, reference Figure 3 Step S30 can be implemented in the following way:

[0064] S301: Extract the maximum predicted pressure value and the minimum predicted pressure value from the pressure prediction values.

[0065] In this embodiment, the maximum predicted pressure value refers to the maximum value among the predicted pressure values ​​within a preset future time period, while the minimum predicted pressure value is the minimum value. By extracting these two values, the range of pressure prediction can be understood, providing a basis for subsequent calculation of pressure fluctuation amplitude.

[0066] In the robot's operating environment, predicted pressure values ​​reflect potential pressure changes over a future period. The maximum and minimum predicted pressure values ​​represent the upper and lower limits of pressure variation, and they are crucial for assessing the environment's pressure stability and explosion-proof requirements. For example, a high maximum predicted pressure value indicates a potentially significant risk of pressure fluctuations, necessitating more stringent explosion-proof measures.

[0067] In one embodiment, the pressure prediction values ​​can be sorted to identify the maximum and minimum values. Alternatively, data processing algorithms, such as traversal algorithms, can be used to directly filter out the maximum and minimum pressure values ​​from the predictions. By accurately extracting the maximum and minimum predicted pressure values, a more comprehensive understanding of pressure changes can be achieved, providing accurate data support for subsequent calculations.

[0068] S302: Calculate the difference between the maximum predicted pressure value and the minimum predicted pressure value, and use it as the pressure fluctuation amplitude.

[0069] In this embodiment of the application, the pressure fluctuation amplitude refers to the difference between the maximum predicted pressure value and the minimum predicted pressure value, which reflects the severity of pressure changes within a preset time period in the future.

[0070] Pressure fluctuation amplitude is an important indicator for assessing environmental pressure stability. Larger pressure fluctuation amplitudes indicate more drastic pressure changes, which may threaten the safe operation of robots; smaller pressure fluctuation amplitudes indicate relatively stable pressure, making explosion-proof control relatively easier. Calculating the pressure fluctuation amplitude provides an important basis for subsequent calculations of dynamic explosion-proof thresholds.

[0071] In one embodiment, the pressure fluctuation amplitude is obtained by directly subtracting the minimum predicted pressure value from the maximum predicted pressure value. This value is then recorded for subsequent calculations. Accurate calculation of the pressure fluctuation amplitude allows for a better understanding of the severity of pressure changes, providing a reference for developing appropriate explosion-proof strategies.

[0072] S303: Based on the pressure fluctuation amplitude and the preset safety margin coefficient, calculate the initial explosion-proof threshold of the dynamic explosion-proof threshold.

[0073] In this embodiment, the safety margin factor is a pre-set coefficient used to account for potential uncertainties and safety margins. The initial explosion-proof threshold is a preliminary explosion-proof threshold calculated based on the pressure fluctuation amplitude and the safety margin factor.

[0074] In practical applications, due to the inherent errors and uncertainties in pressure prediction, a safety margin factor needs to be set to ensure the reliability of the explosion-proof threshold. The size of the safety margin factor affects the calculation result of the initial explosion-proof threshold. A larger safety margin factor results in a higher initial explosion-proof threshold, providing more adequate safety assurance; a smaller safety margin factor results in a relatively lower initial explosion-proof threshold, which may improve production efficiency to some extent, but could reduce the level of safety.

[0075] In one embodiment, the initial explosion-proof threshold is obtained by multiplying the pressure fluctuation amplitude by a safety margin factor and adding a reference pressure value (such as the average pressure during normal operation). Calculating the initial explosion-proof threshold in this way allows for full consideration of safety factors while taking pressure fluctuations into account, providing a reasonable basis for subsequent adjustments.

[0076] S304: The initial explosion-proof threshold is corrected based on the statistical characteristics of historical environmental pressure data to obtain the final dynamic explosion-proof threshold.

[0077] In this embodiment, the statistical characteristics of historical environmental pressure data refer to features obtained from statistical analysis of pressure data over a past period, such as standard deviation and average value. By utilizing these statistical characteristics to correct the initial explosion-proof threshold, the final dynamic explosion-proof threshold can more accurately reflect the actual explosion-proof requirements.

[0078] During robot operation, historical environmental pressure data contains patterns and characteristics of pressure changes. By analyzing the statistical properties of this data, hidden information can be discovered, such as periodic pressure variations and abnormal fluctuations. Using this information to correct the initial explosion-proof threshold can improve the threshold's adaptability and accuracy.

[0079] In one embodiment, pressure change trend data within multiple time windows are extracted from historical environmental pressure data, and the standard deviation and average value of the data within each time window are calculated. Then, a correction coefficient is calculated based on the standard deviation and average value, and this correction coefficient is multiplied by the initial explosion-proof threshold to obtain the corrected explosion-proof threshold. During the correction process, if the correction coefficient is less than a preset lower limit, it is set as the lower limit. By correcting the initial explosion-proof threshold in this way, the final dynamic explosion-proof threshold can be made more consistent with actual conditions, improving the robot's explosion-proof safety.

[0080] In one embodiment, reference Figure 4 Step S304 can be implemented in the following way:

[0081] S3041: Extract pressure change trend data within multiple time windows from historical environmental pressure data.

[0082] In this embodiment, a time window refers to a period of time in which historical environmental pressure data is divided according to a certain time length. Pressure change trend data refers to the changes in pressure data within each time window, such as rising, falling, or remaining stable.

[0083] Historical environmental pressure data records pressure changes over a past period. By dividing the data into time windows and extracting pressure change trends, a more detailed analysis of pressure variation patterns can be achieved. For example, different time windows may correspond to different stages of the production process, resulting in different pressure change trends. Studying this trend data can provide a more accurate basis for revising dynamic explosion-proof thresholds.

[0084] In one embodiment, the length of the time window can be determined based on the production cycle or other relevant factors. Then, a sliding window process is applied to historical environmental pressure data to extract pressure data within each time window and analyze its changing trends. Data analysis methods, such as trend line fitting and difference analysis, can be used to determine the pressure change trend. By accurately extracting pressure change trend data within multiple time windows, a deeper understanding of the pressure change patterns can be achieved, providing more detailed information for subsequent calculations.

[0085] S3042: Calculate the standard deviation and mean of the pressure change trend data within each time window.

[0086] In this embodiment, the standard deviation is a statistic that measures the dispersion of data, reflecting the fluctuation of pressure change trend data relative to the mean. The mean is the arithmetic mean of the data, representing the average level of pressure change.

[0087] Standard deviation and mean are important statistical indicators for describing the characteristics of pressure change trends. A larger standard deviation indicates more drastic pressure changes and higher data dispersion; a smaller standard deviation indicates relatively stable pressure changes. The mean reflects the overall pressure level and is important for determining the normal range of pressure.

[0088] In one embodiment, statistical formulas can be used to calculate the standard deviation and mean of pressure change trend data within each time window. First, the mean of the data is calculated. Then, the sum of squared differences between each data point and the mean is calculated, divided by the number of data points, and finally the square root is taken to obtain the standard deviation. By calculating the standard deviation and mean, the characteristics of the pressure change trend data can be quantified, providing a basis for calculating the correction factor.

[0089] S3043: Calculate a correction factor based on the standard deviation and the mean, wherein the correction factor is a positive number less than or equal to one.

[0090] In this embodiment, the correction coefficient is used to correct the initial explosion-proof threshold, and its value ranges from a positive number less than or equal to one. It comprehensively considers the standard deviation and average value of historical environmental pressure data, reflecting the stability and deviation of pressure changes, so as to adjust the initial explosion-proof threshold to better reflect the actual situation.

[0091] The standard deviation reflects the degree of fluctuation in pressure data, while the average value represents the average pressure level. A large standard deviation indicates severe pressure fluctuations, requiring a significant adjustment to the initial explosion-proof threshold. Conversely, a small standard deviation indicates relatively stable pressure, allowing for a smaller adjustment. The average value also affects the calculation of the correction factor; if the average value deviates from the normal range, the initial explosion-proof threshold also needs to be corrected.

[0092] In one embodiment, a correction factor can be calculated according to a preset formula. For example, the correction factor is equal to a base factor minus an adjustment value related to the standard deviation and the mean. The base factor can be determined based on experience or experiments, while the adjustment value is calculated based on the specific values ​​of the standard deviation and the mean. The correction factor calculated in this way can reasonably reflect the characteristics of historical pressure data, providing an accurate basis for subsequent adjustments to the initial explosion-proof threshold. By calculating the correction factor, the dynamic explosion-proof threshold can more accurately adapt to actual pressure changes, improving the effectiveness of explosion-proof measures.

[0093] S3044: Calculate a correction factor based on the standard deviation and the mean, wherein the correction factor is a positive number less than or equal to one.

[0094] In this embodiment, multiplying the correction coefficient by the initial explosion-proof threshold is a specific method for correcting the initial explosion-proof threshold. Through this multiplication operation, the initial explosion-proof threshold can be adjusted based on the statistical characteristics of historical environmental pressure data, making it more consistent with actual explosion-proof requirements.

[0095] The correction factor reflects the characteristics of historical pressure data. When the correction factor is less than 1, it means that the initial explosion-proof threshold needs to be appropriately reduced; when the correction factor is close to 1, it indicates that the initial explosion-proof threshold is reasonable and only a small adjustment is needed. The corrected explosion-proof threshold obtained in this way can better adapt to the actual situation of pressure changes and avoid problems caused by the initial explosion-proof threshold being too high or too low.

[0096] In one embodiment, the calculated correction coefficient is directly multiplied by the initial explosion-proof threshold to obtain the corrected explosion-proof threshold. This simple and direct calculation method can quickly and accurately correct the initial explosion-proof threshold, providing a reliable threshold standard for subsequent risk assessment and explosion-proof control. By multiplying the correction coefficient by the initial explosion-proof threshold, the dynamic explosion-proof threshold becomes more scientific and reasonable, improving the explosion-proof safety of the robot under different pressure environments.

[0097] S3045: During the correction process, if the correction coefficient is less than a preset lower limit, then the correction coefficient is set to the lower limit.

[0098] In this embodiment, the preset lower limit is a threshold set to ensure that the correction coefficient is not too small. During the correction process, if the calculated correction coefficient is less than this lower limit, it is set as the lower limit to ensure that the corrected explosion-proof threshold is not too low and to ensure the safe operation of the robot.

[0099] Setting a lower limit is to prevent the correction coefficient from becoming too small due to abnormal fluctuations in historical pressure data or calculation errors, thus making the corrected explosion-proof threshold too low and unable to meet actual explosion-proof requirements. By limiting the minimum value of the correction coefficient, the dynamic explosion-proof threshold can be ensured to be within a reasonable range, avoiding safety hazards caused by an excessively low threshold.

[0100] In one embodiment, after calculating the correction coefficient, it is compared with a preset lower limit value. If the correction coefficient is less than the lower limit value, the correction coefficient is replaced with the lower limit value, and subsequent multiplication operations are performed to obtain the corrected explosion-proof threshold. This method ensures that the corrected explosion-proof threshold has a certain level of safety and reliability, improving the robot's explosion-proof capability in complex pressure environments.

[0101] In one embodiment, reference Figure 5 The method in this application embodiment may further include the process of adjusting the modified explosion-proof threshold, as follows:

[0102] S60: Acquire multi-dimensional environmental factor data in the robot's operating environment, including temperature, humidity, and gas concentration.

[0103] In this embodiment, multi-dimensional environmental factor data refers to data on other relevant environmental factors in the robot's operating environment besides pressure data, including temperature, humidity, and gas concentration. These factors affect environmental pressure, thereby influencing the setting of the explosion-proof threshold.

[0104] Temperature, humidity, and gas concentration are crucial factors affecting the safety of flammable and explosive environments. Changes in temperature can cause gases to expand or contract, thus affecting pressure; humidity levels can influence the chemical reactivity of gases, thereby affecting the likelihood of an explosion; and gas concentration directly relates to the degree of explosion hazard. Obtaining data on these multi-dimensional environmental factors provides a more comprehensive understanding of the robot's operating environment, offering a basis for accurately adjusting dynamic explosion-proof thresholds.

[0105] In one embodiment, appropriate sensors, such as temperature sensors, humidity sensors, and gas concentration sensors, can be installed around the robot to collect environmental factor data in real time. The sensors transmit the collected data to the control platform, which stores and processes the data. By acquiring multi-dimensional environmental factor data, the impact of environmental factors on pressure and explosion protection can be considered more comprehensively, improving the accuracy and adaptability of dynamic explosion protection thresholds.

[0106] S61: Conduct correlation analysis between multi-dimensional environmental factor data and historical stress data to extract the influence weight of each environmental factor on stress changes.

[0107] In this embodiment, correlation analysis refers to studying the relationship between multi-dimensional environmental factor data and historical stress data to identify the degree of influence of each environmental factor on stress changes. Influence weight refers to the relative importance of each environmental factor to stress changes.

[0108] Different environmental factors may have varying degrees of impact on pressure changes. For example, in some environments, temperature changes may have a significant impact on pressure, while in others, changes in gas concentration may be more critical. By extracting influence weights through correlation analysis, the role of each environmental factor in pressure changes can be clarified, providing a basis for subsequent calculations of the contribution of each environmental factor to the dynamic explosion-proof threshold.

[0109] In one embodiment, statistical analysis methods, such as correlation analysis and regression analysis, can be used to process multi-dimensional environmental factor data and historical pressure data. By calculating the correlation coefficient or regression coefficient between each environmental factor and the pressure data, their influence weight on pressure changes can be determined. This allows for a quantitative description of the degree of influence of each environmental factor on pressure, providing a scientific basis for accurately adjusting the dynamic explosion-proof threshold. Extracting influence weights through correlation analysis enables a more precise consideration of the impact of each environmental factor on pressure and explosion-proof, improving the rationality of the dynamic explosion-proof threshold.

[0110] S62: Calculate the contribution of each environmental factor to the dynamic explosion-proof threshold based on the influence weight.

[0111] In this embodiment, the contribution value refers to the degree of influence of each environmental factor on the dynamic explosion-proof threshold according to its influence weight. By calculating the contribution value, the specific adjustment amount of each environmental factor to the dynamic explosion-proof threshold can be determined.

[0112] The influence weights reflect the relative importance of each environmental factor to pressure changes. Calculating the contribution value based on the influence weights allows the impact of each environmental factor to be quantified and incorporated into the adjustment of the dynamic explosion-proof threshold. For example, if the influence weight of a certain environmental factor is large, its contribution to the dynamic explosion-proof threshold will also be correspondingly large.

[0113] In one embodiment, the measured value of each environmental factor can be multiplied by its corresponding influence weight to obtain the contribution value of that environmental factor to the dynamic explosion-proof threshold. For example, if the influence weight of temperature is 0.3, and the currently measured temperature is 30°C, then the contribution value of temperature to the dynamic explosion-proof threshold is 30 × 0.3 = 9 (assuming the unit is a certain pressure adjustment unit). By calculating the contribution value of each environmental factor, the influence of each environmental factor on the dynamic explosion-proof threshold can be considered more accurately, making the threshold adjustment more scientific and reasonable.

[0114] S63: Adjust the modified explosion-proof threshold based on the contribution value to the dynamic explosion-proof threshold.

[0115] In this embodiment, adjusting the modified explosion-proof threshold based on the contribution value is to further improve the dynamic explosion-proof threshold, ensuring it fully considers the impact of multi-dimensional environmental factors. This adjustment makes the dynamic explosion-proof threshold more consistent with actual environmental conditions, thereby improving the robot's explosion-proof safety.

[0116] The contribution values ​​of each environmental factor reflect their specific impact on pressure and explosion protection. By taking these contribution values ​​into account in the revised explosion protection threshold, the threshold can be adjusted more precisely. For example, if the contribution value of an environmental factor is positive, it means that the factor will increase the pressure, and the dynamic explosion protection threshold needs to be increased accordingly; if the contribution value is negative, the threshold needs to be decreased.

[0117] In one embodiment, the contributions of each environmental factor are summed to obtain the total adjustment amount. This adjustment amount is then added to or subtracted from the corrected explosion-proof threshold to obtain the final adjusted dynamic explosion-proof threshold. Adjusting the corrected explosion-proof threshold in this way allows the dynamic explosion-proof threshold to more accurately adapt to the actual environment, improving the robot's explosion-proof capability in complex environments.

[0118] In one embodiment, reference Figure 6 Step S40 may specifically include the following steps:

[0119] S401: Obtain the difference between the actual value of the pressure data and the dynamic explosion-proof threshold, the difference being used to characterize the degree to which the current pressure state deviates from the safe range.

[0120] In this embodiment, the actual pressure data refers to the pressure value in the robot's operating environment collected in real time by a high-precision pressure sensor. The dynamic explosion-proof threshold is the explosion-proof safety limit that changes over time, calculated based on the pressure prediction value and related corrections. Obtaining the difference between these two values ​​allows for the quantification of the deviation between the current pressure state and the safe range.

[0121] The magnitude of the difference directly reflects whether the current pressure is within a safe range and the degree of deviation. If the difference is positive and large, it indicates that the actual pressure exceeds the dynamic explosion-proof threshold and is in a dangerous state; if the difference is negative and large in absolute value, it indicates that the actual pressure is far below the dynamic explosion-proof threshold and there may be a problem of excessively low pressure; if the difference is close to zero, it indicates that the current pressure state is relatively safe.

[0122] In one embodiment, the difference is obtained by directly subtracting the dynamic explosion-proof threshold from the actual pressure data. This calculation process is simple and direct, and can quickly and accurately determine the degree to which the current pressure deviates from the safe range. By obtaining the difference, a key quantitative indicator is provided for subsequent risk level determination, which helps to detect abnormal pressure situations in a timely manner and ensure the safe operation of the robot.

[0123] S402: Map the difference to a preset risk level range, which includes a low-risk range, a medium-risk range, and a high-risk range.

[0124] In this embodiment, the preset risk level ranges are different difference ranges pre-defined based on experience and safety standards, corresponding to low risk, medium risk, and high risk, respectively. Mapping the differences to these ranges allows for a direct assessment of the risk level of the current pressure state.

[0125] The risk level ranges are used to classify and manage pressure deviations of varying degrees. A low-risk range indicates a small deviation from the safe range with minimal impact on the robot and its surroundings; a medium-risk range indicates a moderate deviation requiring appropriate attention; and a high-risk range indicates a significant deviation that could lead to dangerous situations such as explosions, requiring immediate action.

[0126] In one embodiment, mapping can be performed by comparing the difference with the upper and lower limits of a risk level range. For example, if the difference is less than the upper limit of the low-risk range, it is mapped to the low-risk range; if the difference is greater than the upper limit of the low-risk range but less than the upper limit of the medium-risk range, it is mapped to the medium-risk range; and if the difference is greater than the upper limit of the medium-risk range, it is mapped to the high-risk range. By mapping the difference to a preset risk level range, the risk level of the current pressure state can be quickly and accurately determined, providing a basis for decision-making in subsequent explosion-proof control.

[0127] S403: If the difference is in the low-risk range, the current risk level is determined to be Level 1; if the difference is in the medium-risk range, the current risk level is determined to be Level 2; if the difference is in the high-risk range, the current risk level is determined to be Level 3.

[0128] In this embodiment, the risk level is determined based on the risk level range in which the difference falls, in order to manage different degrees of pressure deviation in a tiered manner. Level 1 risk represents low risk, Level 2 risk represents medium risk, and Level 3 risk represents high risk.

[0129] Different risk levels correspond to different explosion-proof control measures. For Level 1 risk, the intensity of explosion-proof control can be appropriately reduced to improve production efficiency; for Level 2 risk, pressure monitoring and control need to be strengthened; for Level 3 risk, emergency explosion-proof measures must be taken immediately to ensure the safety of the robot and the surrounding environment.

[0130] In one embodiment, after mapping the difference to the risk level range, the current risk level is directly determined based on the mapping result. This explicit classification method helps to quickly respond to different levels of pressure risks, rationally allocate explosion-proof resources, and improve the safety and operational efficiency of robots in flammable and explosive environments. By accurately determining the risk level, targeted explosion-proof control strategies can be adopted to ensure the safety and stability of production.

[0131] In one embodiment, reference Figure 7 Step S50 may specifically include the following steps:

[0132] S501: Extract the corresponding explosion-proof control parameters from the preset explosion-proof strategy library according to the current risk level. The explosion-proof control parameters include pressure release rate, pressure adjustment cycle and pressure upper limit value.

[0133] In this embodiment, the preset explosion-proof strategy library is a database that pre-stores explosion-proof control strategies and parameters corresponding to different risk levels. Explosion-proof control parameters are specific parameters used to control equipment such as robot pressure regulating devices, including pressure release rate, pressure regulation cycle, and pressure upper limit.

[0134] Different risk levels correspond to different explosion-proof requirements, necessitating different explosion-proof control measures. The parameters stored in the explosion-proof strategy library are determined based on experience and experiments, ensuring the safe operation of robots under various risk conditions. For example, at high-risk levels, a faster pressure release rate and a shorter pressure regulation cycle are needed to quickly reduce pressure and avoid the risk of explosion.

[0135] In one embodiment, based on the currently determined risk level, the corresponding record is searched in the explosion-proof strategy library, and parameters such as pressure release rate, pressure adjustment cycle, and pressure upper limit are extracted. This method of extracting parameters from a preset library is simple and efficient, and can quickly provide accurate parameter basis for the subsequent generation of explosion-proof control commands. By extracting the corresponding explosion-proof control parameters according to the risk level, explosion-proof control strategies can be formulated in a targeted manner, improving the explosion-proof safety of the robot.

[0136] S502: Generate explosion-proof control instructions based on the explosion-proof control parameters. The explosion-proof control instructions include specific execution actions and execution times.

[0137] In this embodiment, the explosion-proof control command is a specific operation command generated based on the extracted explosion-proof control parameters, used to control equipment such as the robot's pressure regulating device. The command includes specific execution actions and execution times, clearly defining the equipment's operation mode and timing.

[0138] Explosion-proof control parameters define the operating parameters of the equipment, while explosion-proof control commands translate these parameters into specific operating procedures. The execution action specifies the operation the equipment needs to perform, such as opening or closing valves or adjusting pump speed; the execution time determines the specific moment or time interval of these operations.

[0139] In one embodiment, based on the pressure release rate, pressure regulation cycle, and pressure upper limit in the explosion-proof control parameters, combined with the robot's equipment characteristics and operating procedures, specific execution actions and execution times are generated. For example, if the pressure release rate is required to release a certain amount of pressure per minute, then an instruction to open the valve to release pressure at the beginning of each pressure regulation cycle can be generated, specifying the specific opening and closing times. By generating explosion-proof control instructions containing specific execution actions and execution times based on the explosion-proof control parameters, the robot's pressure regulation equipment can be accurately controlled, achieving effective explosion-proof control.

[0140] S503: The explosion-proof control command is sent to the robot, and the robot adjusts the working state of the pressure regulating device according to the explosion-proof control command.

[0141] In this embodiment, sending explosion-proof control commands to the robot is a key step in implementing the explosion-proof control strategy. Upon receiving the commands, the robot adjusts the operating state of the pressure regulating device according to the execution actions and times specified in the commands, thereby controlling the pressure and ensuring safety.

[0142] Pressure regulating devices, such as valves and pumps, are equipment used by robots to regulate environmental pressure. Explosion-proof control commands specify the operating methods and timing for these devices. By executing these commands, the robot can adjust the pressure in a timely manner to maintain it within a safe range.

[0143] In one embodiment, the control platform sends explosion-proof control commands to the robot via a wireless network. Upon receiving the commands, the robot's control system parses the execution actions and timings, converting them into control signals for the pressure regulating device. For example, if the command requires opening a valve to release pressure, the robot's control system sends a signal to the valve actuator to open the valve. By sending explosion-proof control commands to the robot and instructing it to adjust the operating state of the pressure regulating device, real-time explosion-proof control of the robot can be achieved, improving the robot's safety in flammable and explosive environments.

[0144] In one embodiment, reference Figure 8 Step S20 can be implemented in the following way:

[0145] S201: Normalize the pressure data to obtain standardized pressure data.

[0146] In this embodiment, normalization maps the pressure data to a specific interval, typically [0,1] or [-1, 1], to eliminate differences in the dimensions and scales of the data, resulting in standardized pressure data. Standardized pressure data has a uniform scale within this specific interval, facilitating subsequent processing and analysis.

[0147] The raw values ​​of stress data may have different dimensions and ranges, which can affect the training performance and prediction accuracy of Long Short-Term Memory (LSTM) network models. Normalization ensures that data have equal importance in the model, avoiding model training instability caused by differences in data scale.

[0148] In one embodiment, a min-max normalization method can be used, which involves subtracting the minimum stress value from the stress data and then dividing by the difference between the maximum and minimum values. The formula is: Standardized stress data = (Original stress data - Minimum stress data) / (Maximum stress data - Minimum stress data). This method is simple and intuitive, and can effectively normalize stress data to the [0, 1] interval. By normalizing the stress data, standardized stress data can be obtained, which can improve the training efficiency and prediction accuracy of the Long Short-Term Memory (LSTM) network model.

[0149] S202: Input the standardized stress data into a long short-term memory network model, which includes an input layer, a hidden layer, and an output layer.

[0150] In this embodiment, the Long Short-Term Memory (LSTM) network model is a special type of recurrent neural network, consisting of an input layer, a hidden layer, and an output layer. The input layer receives standardized stress data, the hidden layer performs feature learning and processing on the data, and the output layer generates predicted stress values ​​for a predetermined future time period.

[0151] Standardized stress data, after normalization, has a uniform scale and range, making it suitable as input for a Long Short-Term Memory (LSTM) network model. The model's three-layer structure has a clear division of labor: the input layer introduces the data into the model, the hidden layer uses LSTM units to perform deep processing on the data, uncovering long-term dependencies in the stress data, and the output layer transforms the processing results into predicted stress values.

[0152] In one embodiment, standardized stress data is sequentially input into the input layer in chronological order. The input layer then passes the data to the hidden layer, where LSTM units process and learn the data. Finally, the output layer generates a stress prediction value based on the output of the hidden layer. By inputting standardized stress data into the Long Short-Term Memory (LSTM) network model, the model's advantages can be fully utilized to accurately predict future stress changes.

[0153] S203: Perform feature learning on the standardized stress data through the hidden layer to extract long-term dependencies in the time series.

[0154] In this embodiment, the hidden layer is the core component of the Long Short-Term Memory (LSTM) network model. It uses LSTM units to learn features from standardized stress data and extract long-term dependencies in the time series. Long-term dependencies refer to the associations and influences of stress data at different points in time, such as periodic changes or trend changes.

[0155] Changes in stress data often exhibit certain patterns and long-term correlations. Traditional neural networks struggle to handle these long-term dependencies, while the LSTM units of Long Short-Term Memory (LSTM) networks, through their input, forget, and output gates, can effectively capture and learn these relationships. By extracting these long-term dependencies, future stress trends can be predicted more accurately.

[0156] In one embodiment, the LSTM units in the hidden layer receive standardized stress data. An input gate controls the weight of new input data, a forget gate controls the retention of historical data, and an output gate controls the output of the current unit state. At each time step, the LSTM unit updates its own state based on the input data and historical states, thereby learning the long-term dependencies of the stress data. By performing feature learning on the standardized stress data through the hidden layer and extracting long-term dependencies in the time series, the accuracy and reliability of stress prediction can be improved.

[0157] In one embodiment, reference Figure 9 Step S203 can be implemented in the following way:

[0158] S2031: Input the standardized pressure variable data into the hidden layer, which includes multiple LSTM units, each of which contains an input gate, a forget gate, and an output gate.

[0159] In this embodiment, the standardized stress variable data is data with a uniform scale and range after normalization, making it suitable as input to the hidden layer of a Long Short-Term Memory (LSTM) network. The hidden layer consists of multiple LSTM units, which are the core structure of the LSTM network. Each LSTM unit contains an input gate, a forget gate, and an output gate. These three gating mechanisms work together to enable the LSTM unit to effectively handle long-term dependencies in time-series data.

[0160] The input gate controls the weights of new input data, determining which new information can enter the cell state; the forget gate controls the retention of historical data, determining which historical information needs to be forgotten; and the output gate controls the output of the current cell state, determining which information can be passed to the next layer or the next time step as the output of the current cell. Through the control of these three gates, the LSTM cell can selectively remember and forget information, thereby better capturing the changing patterns of stress data over time.

[0161] In one embodiment, standardized stress variable data is sequentially input into each LSTM unit of the hidden layer in chronological order. Upon receiving the input data, each LSTM unit performs corresponding calculations and decisions based on the current input and historical states. For example, the input gate calculates a weight vector based on the input data and the previous hidden state, determining which parts of the new input data can enter the unit state. By inputting standardized stress variable data into a hidden layer containing multiple LSTM units, a foundation is laid for subsequent extraction of long-term dependencies in the time series.

[0162] S2032: The input gate controls the weight of new input data, the forget gate controls the retention of historical data, and the output gate controls the output of the current unit state.

[0163] In this embodiment, the input gate, forget gate, and output gate are components of the LSTM unit, each undertaking different responsibilities to jointly achieve effective data processing and capture of long-term dependencies. The input gate determines which parts of the new input data can be included in the unit state by calculating weights, the forget gate determines which information in the historical data needs to be retained or forgotten based on the calculation results, and the output gate controls which parts of the current unit state can be passed out as output.

[0164] When processing time series data, new input data may contain information that significantly impacts prediction, but it may also contain noise or irrelevant information. The input gate, by calculating weights, can filter out valuable information and incorporate it into the cell state. Meanwhile, as time progresses, some information in historical data may become irrelevant; the forget gate can promptly remove this information, preventing information overload. The output gate, based on the current cell state and task requirements, selectively outputs information helpful for subsequent predictions.

[0165] In one embodiment, the input gate calculates a weight vector ranging from [0, 1] using an activation function (such as the sigmoid function) based on the current input data and the hidden state of the previous time step. Each element in the weight vector represents the weight of the corresponding input data; a value close to 1 indicates that the data can be fully incorporated into the cell state, while a value close to 0 indicates that it is ignored. The forget gate and output gate are calculated in a similar manner. For example, the forget gate calculates a forget vector using the sigmoid function, determining the degree to which each element in the historical cell state is retained. Through the coordinated control of the input gate, forget gate, and output gate, the LSTM cell can effectively process the time-series information of stress data and extract long-term dependencies.

[0166] S2033: At each time step, calculate the state update value of the hidden layer, which is obtained by weighted sum of the current input data and the historical state.

[0167] In this embodiment, when processing time series data, each time step corresponds to a new input data. The state update value of the hidden layer reflects the result of updating the cell state based on the new input data and historical states at the current time step. Calculating the state update value by weighting the current input data and historical states fully considers the combined influence of historical and new information, thereby better capturing long-term dependencies in the time series.

[0168] At each time step, after the LSTM unit receives new input data, the input gate filters and weights it, while the forget gate selectively retains historical states. Then, the processed new input data and the retained historical states are weighted and summed to obtain the new unit state. This process allows the unit state to be continuously updated over time while preserving important historical information.

[0169] In one embodiment, it is assumed that the current input data is The hidden state at the previous moment was , unit state The weight vector calculated by the input gate is: The forgetting vector calculated by the forgetting gate is Then the new cell state It can be calculated using the following formula:

[0170] C t = f t ⊙ C t - 1 + i t ⊙ tanh ( W c [ x t , h t - 1 ]+ b c )

[0171] ,in This indicates element-wise multiplication. and These are learnable parameters. By calculating the hidden layer's state update values ​​in this way, the LSTM cells can dynamically update their state at each time step, adapting to changes in the stress data.

[0172] S2034: Use the state update values ​​of the hidden layer to extract long-term dependencies in the time series, which are used to capture periodic fluctuations and trend changes in stress variable data.

[0173] In this embodiment, the state update values ​​of the hidden layer contain all the important information from historical data to the current time step. By analyzing and processing these state update values, long-term dependencies in the time series can be extracted. Stress variable data often exhibits periodic fluctuations and trends in time series data. For example, daily stress may change periodically with the regularity of production activities, or stress may show an upward or downward trend over a period of time. Extracting these long-term dependencies helps to accurately predict future stress changes.

[0174] The state update process of an LSTM cell is a continuous accumulation and updating of information. The state update value at each time step is influenced by historical states and current inputs. By analyzing these state update values, correlations and patterns in stress data at different time points can be discovered. For example, periodic fluctuations may manifest as similar patterns in state update values ​​within a certain time interval, while trend changes may manifest as monotonically increasing or decreasing state update values.

[0175] In one embodiment, feature extraction and analysis of the hidden layer's state update values, such as using time series analysis methods or machine learning algorithms, can identify periodic fluctuations and trend changes in stress variable data. For example, an autoregressive integral moving average (ARIMA) model can be used to model the state update values ​​and extract their periodic and trend features. By utilizing the hidden layer's state update values ​​to extract long-term dependencies, a better understanding of the changing patterns in stress data can be achieved, improving the accuracy of stress prediction.

[0176] S2035: Pass the output of the hidden layer to the next layer of the network. The output contains the learning results of the time series features and is used to generate stress prediction values.

[0177] In this embodiment, the output of the hidden layer is data processed by the LSTM unit and includes the learning results of time-series features. Passing this output to the next layer (usually the output layer) allows the generation of stress prediction values ​​for a predetermined future time period using these learning results. By processing and analyzing standardized stress variable data, the hidden layer extracts long-term dependencies and patterns of change in the stress data. This information is included in the output of the hidden layer, providing crucial information for subsequent stress prediction.

[0178] After receiving the output from the hidden layer, the output layer uses the learned time-series features and a specific mapping function or model to calculate and generate predicted pressure values. Because the hidden layer has already effectively extracted and processed the pressure data, the output layer can more accurately predict future pressure changes based on this information.

[0179] In one embodiment, the output of the hidden layer is The output layer can be transformed by a linear transformation. To generate standardized pressure prediction values, where and These are learnable parameters. Then, the standardized pressure predictions are inversely normalized to obtain predictions within the actual pressure range. By passing the output of the hidden layer to the next layer, the time-series features learned by the hidden layer can be applied to pressure prediction, improving the accuracy and reliability of the predictions.

[0180] S204: Use the output layer to generate a pressure prediction value for a future preset time period, and inversely normalize the pressure prediction value to a prediction result within the actual pressure value range.

[0181] In this embodiment, the output layer is the last layer of the Long Short-Term Memory (LSTM) network model. Its main function is to generate predicted stress values ​​for a predetermined future time period based on the features and long-term dependencies learned by the hidden layers. Since the input data has been normalized, the predicted stress values ​​are also within the normalized range. Therefore, an inverse normalization operation is needed to convert them into predicted results within the actual stress value range for practical application.

[0182] The output layer generates pressure prediction values ​​based on the output of the hidden layer through certain calculations and mapping relationships. These prediction values ​​reflect the possible changes in pressure over a preset time period, but because they are normalized values, they cannot be directly used for actual pressure assessment and control. The denormalization process restores these values ​​to the original pressure data scale, giving them actual physical meaning.

[0183] In one embodiment, the output layer, based on the time-series feature learning results passed from the hidden layer, uses a pre-trained mapping function to generate standardized pressure variable prediction values ​​for a preset future time period. These prediction values ​​are then input to the denormalization module. The denormalization module performs inverse operations based on pre-stored normalization parameters, such as maximum, minimum, and mean values. For example, if the minimum-maximum normalization method is used, the denormalization formula is: Actual pressure prediction value = Standardized pressure prediction value × (Maximum pressure data - Minimum pressure data) + Minimum pressure data. Through this denormalization operation, the standardized pressure variable prediction values ​​are converted into prediction results within the actual pressure value range. The resulting prediction results can be directly used for subsequent dynamic explosion-proof threshold calculations and risk assessments, improving the practicality and operability of the model's prediction results.

[0184] In one embodiment, step S204 can be implemented as follows:

[0185] A: Receive the learning results of time series features from the hidden layer, and generate standardized stress variable prediction values ​​for a future preset time period through the output layer.

[0186] In this embodiment, the output layer receives the learning results of time-series features from the hidden layer. These results contain the long-term dependencies and changing patterns of stress data over time. Based on these learning results, the output layer uses a pre-trained mapping function or model to generate standardized stress variable prediction values ​​for a predetermined future time period. The standardized stress variable prediction values ​​are values ​​within a normalized interval, reflecting the possible future trends in stress.

[0187] The hidden layer processes and analyzes standardized stress variable data using LSTM units, extracting key features and long-term dependencies. The output layer uses this information, combined with its own parameters and structure, to predict future stress. The mapping function of the output layer can be linear or non-linear, depending on the model training and specific design requirements.

[0188] B: Input the predicted value of the standardized pressure variable into the denormalization module. The denormalization module performs inverse operation based on the pre-stored normalization parameters, which include the maximum value, minimum value, and mean value.

[0189] In this embodiment, the main function of the denormalization module is to convert the standardized pressure variable prediction values ​​into prediction results within the actual pressure value range. Since the standardized pressure variable prediction values ​​are normalized values ​​within a specific interval and lack actual physical meaning, a denormalization operation is necessary. The denormalization module performs an inverse operation based on pre-stored normalization parameters, such as maximum, minimum, and mean values, to restore the prediction values ​​to the original pressure data scale.

[0190] Normalization parameters are recorded during the normalization process of the raw pressure data; they reflect the scale and range of the original data. By using these parameters for inverse normalization, it can be ensured that the prediction results have the same physical meaning and scale as the original data.

[0191] In one embodiment, if the min-max normalization method is used, assuming that the maximum value recorded during normalization is max and the minimum value is min, the predicted value of the standardized stress variable is... The inverse normalized predicted actual pressure value It can be calculated using the following formula:

[0192] .

[0193] After receiving the standardized pressure variable prediction value, the inverse normalization module automatically performs the aforementioned inverse operation based on the pre-stored normalization parameters to obtain the prediction result within the actual pressure value range. By inputting the standardized pressure variable prediction value into the inverse normalization module for inverse operation, the prediction result can be directly used for actual pressure assessment and explosion-proof control.

[0194] C: The inverse normalization module converts the standardized pressure variable predictions into predictions within the range of actual pressure values.

[0195] In this embodiment, after the inverse normalization module completes the inverse operation, it successfully converts the predicted value of the standardized pressure variable into a predicted result within the actual pressure value range. This predicted result has practical physical meaning and can be directly used to assess and analyze the pressure in the robot's operating environment, providing accurate data support for subsequent dynamic explosion-proof threshold calculation and risk level determination.

[0196] The transformed actual pressure prediction results reflect the possible values ​​of pressure within a preset time period, with units and scales consistent with the original pressure data. This transformation enables the pressure prediction results based on the Long Short-Term Memory (LSTM) network model to be better applied to real-world scenarios, improving the model's practicality and operability.

[0197] In one embodiment, the denormalization module performs a linear transformation on the predicted values ​​of standardized pressure variables based on pre-stored normalization parameters such as maximum, minimum, and mean values, mapping them from the normalized interval to the range of values ​​in the original pressure data. For example, if the pressure data is mapped to the [0, 1] interval during normalization, the denormalization module will restore the predicted values ​​to the actual pressure values ​​based on the maximum and minimum values ​​of the original data. In this way, the predicted values ​​of standardized pressure variables are converted into prediction results within the actual pressure value range, enabling the prediction results to be directly used for actual pressure management and explosion-proof control decisions, thereby improving the safety and reliability of the robot in flammable and explosive environments.

[0198] Accordingly, to better implement the above methods, this application also provides a robot explosion-proof dynamic threshold control system 80. For example... Figure 10 As shown, the robot explosion-proof dynamic threshold control system 80 includes:

[0199] The acquisition module 801 is used to collect pressure data in the robot's operating environment in real time through a high-precision pressure sensor. The pressure data includes pressure values ​​at multiple time points.

[0200] The prediction processing module 802 is used to input the stress data into a pre-trained long short-term memory network model for time-series prediction processing, and generate stress prediction values ​​for a future preset time period.

[0201] Threshold calculation module 803 is used to calculate a dynamic explosion-proof threshold based on the pressure prediction value, wherein the dynamic explosion-proof threshold changes over time and is associated with the pressure prediction value.

[0202] Risk determination module 804 is used to determine the current risk level based on the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold.

[0203] The control module 805 is used to adjust the explosion-proof control strategy based on the current risk level, generate corresponding explosion-proof control instructions and send them to the robot to achieve dynamic control.

[0204] The implementation details of each module are provided in the preceding method embodiments and will not be repeated here. The technical effects achieved by each module and device are described in the foregoing method embodiments.

[0205] like Figure 11 As shown, this application embodiment also provides a computer device 90, which includes a processor 901 and a memory 902, wherein the memory 902 stores a computer program, and when the computer program is executed by the processor 901, the processor 901 performs the steps of any of the methods described above.

[0206] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0207] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A method for adjusting the dynamic threshold for explosion protection of a robot, characterized in that, Includes the following steps: The robot's operating environment is collected in real time using a high-precision pressure sensor, and the pressure data includes pressure values ​​at multiple time points. The stress data is input into a pre-trained long short-term memory network model for time-series prediction processing to generate stress prediction values ​​for a future preset time period. Extract the maximum and minimum predicted pressure values ​​from the predicted pressure values; Calculate the difference between the maximum predicted pressure value and the minimum predicted pressure value, and use it as the pressure fluctuation amplitude; Based on the pressure fluctuation amplitude and the preset safety margin coefficient, the initial explosion-proof threshold of the dynamic explosion-proof threshold is calculated; Extract pressure change trend data within multiple time windows from historical environmental pressure data; Calculate the standard deviation and mean of the pressure change trend data within each time window; The correction factor is calculated based on the standard deviation and the mean, and the correction factor is a positive number less than or equal to one. Multiply the correction coefficient by the initial explosion-proof threshold to obtain the corrected dynamic explosion-proof threshold; During the correction process, if the correction coefficient is less than the preset lower limit, the correction coefficient is set to the lower limit. The dynamic explosion-proof threshold changes over time and is associated with the pressure prediction value. The current risk level is determined based on the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold. Based on the current risk level, the explosion-proof control strategy is adjusted, corresponding explosion-proof control instructions are generated and sent to the robot to achieve dynamic control.

2. The method according to claim 1, characterized in that, The method further includes: Acquire multi-dimensional environmental factor data of the robot's operating environment, including temperature, humidity, and gas concentration; By correlating multidimensional environmental factor data with historical stress data, the influence weight of each environmental factor on stress changes is extracted. The contribution of each environmental factor to the dynamic explosion-proof threshold is calculated based on its influence weight. The modified explosion-proof threshold is adjusted based on the contribution value.

3. The method according to claim 1, characterized in that, The step of determining the current risk level based on the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold includes: The difference between the actual value of the pressure data and the dynamic explosion-proof threshold is obtained, and the difference is used to characterize the degree to which the current pressure state deviates from the safe range; The difference is mapped to a preset risk level range, which includes a low-risk range, a medium-risk range, and a high-risk range; If the difference is in the low-risk range, the current risk level is determined to be Level 1; if the difference is in the medium-risk range, the current risk level is determined to be Level 2; if the difference is in the high-risk range, the current risk level is determined to be Level 3.

4. The method according to claim 1, characterized in that, The process of adjusting the explosion-proof control strategy based on the current risk level, generating corresponding explosion-proof control commands, and sending them to the robot for dynamic adjustment includes: Based on the current risk level, the corresponding explosion-proof control parameters are extracted from the preset explosion-proof strategy library. The explosion-proof control parameters include pressure release rate, pressure adjustment cycle, and pressure upper limit value. Explosion-proof control commands are generated based on the explosion-proof control parameters. The explosion-proof control commands include specific execution actions and execution times. The explosion-proof control command is sent to the robot, and the robot adjusts the working state of the pressure regulating device according to the explosion-proof control command.

5. The method according to claim 1, characterized in that, The step of inputting the stress data into a pre-trained long short-term memory network model for time-series prediction processing to generate predicted stress values ​​for a future preset time period includes: The pressure data is normalized to obtain standardized pressure data; The standardized stress data is input into a long short-term memory network model, which includes an input layer, a hidden layer, and an output layer. The hidden layer is used to learn features from the standardized stress data and extract long-term dependencies in the time series. The output layer is used to generate a pressure prediction value for a future preset time period, and the pressure prediction value is inversely normalized to a prediction result within the actual pressure value range.

6. The method according to claim 5, characterized in that, The step of performing feature learning on the standardized stress data based on the hidden layer to extract long-term dependencies in the time series includes: The standardized pressure data is input into the hidden layer, which includes multiple LSTM units, each of which contains an input gate, a forget gate, and an output gate. The input gate controls the weight of new input data, the forget gate controls the degree of retention of historical data, and the output gate controls the output of the current unit state. At each time step, the state update value of the hidden layer is calculated, which is obtained by weighted sum of the current input data and the historical state; The state update values ​​of the hidden layer are used to extract long-term dependencies in the time series, which are used to capture periodic fluctuations and trend changes in stress variable data; The output of the hidden layer is passed to the next layer of the network. The output contains the learning results of the time series features and is used to generate stress prediction values.

7. The method according to claim 6, characterized in that, The step of generating a pressure prediction value for a preset time period using the output layer and then inversely normalizing the pressure prediction value to a prediction result within the actual pressure value range includes: The learning results of time series features are received from the hidden layer, and the standardized stress variable prediction values ​​for a future preset time period are generated through the output layer. The standardized pressure variable prediction values ​​are input into the inverse normalization module, which performs inverse operations based on pre-stored normalization parameters, including the maximum value, minimum value, and mean value. The inverse normalization module converts the standardized pressure variable predictions into predictions within the range of actual pressure values.

8. A robot explosion-proof dynamic threshold control system, characterized in that, The system includes: The acquisition module is used to collect pressure data in the robot's operating environment in real time through a high-precision pressure sensor. The pressure data includes pressure values ​​at multiple time points. The prediction processing module is used to input the stress data into a pre-trained long short-term memory network model for time-series prediction processing, and generate stress prediction values ​​for a future preset time period. The threshold calculation module is used to extract the maximum and minimum predicted pressure values ​​from the predicted pressure values; calculate the difference between the maximum and minimum predicted pressure values ​​and use it as the pressure fluctuation amplitude; calculate the initial explosion-proof threshold of the dynamic explosion-proof threshold based on the pressure fluctuation amplitude and a preset safety margin coefficient; extract pressure change trend data within multiple time windows from historical environmental pressure data; calculate the standard deviation and average value of the pressure change trend data within each time window; calculate a correction coefficient based on the standard deviation and average value, wherein the correction coefficient is a positive number less than or equal to one; multiply the correction coefficient by the initial explosion-proof threshold to obtain the corrected dynamic explosion-proof threshold; during the correction process, if the correction coefficient is less than a preset lower limit value, then the correction coefficient is set to the lower limit value, and the dynamic explosion-proof threshold changes over time and is associated with the predicted pressure values; The risk determination module is used to determine the current risk level based on the relationship between the actual value of the pressure data and the dynamic explosion-proof threshold. The control module is used to adjust the explosion-proof control strategy based on the current risk level, generate corresponding explosion-proof control commands, and send them to the robot to achieve dynamic control.

Images