A clean cold operation room clean environment control system

Abstract

This application relates to the field of intelligent control technology. Specifically, it discloses a clean environment control system for a clean cold operation room. By collecting the air particulate matter concentration and the pressure data before and after the HEPA filter in the operation room in real time, and using the time series analysis technology based on deep learning to perform time series modeling on the air particulate matter concentration data and the pressure drop data before and after the filter, it captures the changes in the clean environment and the filter load state in the operation room. Furthermore, through the time series correlation response analysis of the air particulate matter concentration and the filter pressure drop, it excavates the potential correlation patterns between the air cleanliness and the filter working state in the operation room, so as to dynamically adjust the air volume of the air supply fan. In this way, the adaptive control of the air cleanliness in the clean operation room can be realized, while optimizing the use efficiency of the HEPA filter, reducing unnecessary energy consumption and the filter replacement frequency.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology, and more specifically, to a clean environment control system for a clean cold operation room. Background Art

[0002] In key industries such as high-tech manufacturing, biomedicine, and precision electronics assembly, clean cold operation rooms are crucial for maintaining a high level of cleanliness and suitable temperature in the production environment. Specifically, these industries require extremely high air quality during the production process to ensure the stability of the production process and the high quality of products. If the concentration of particles in the air (such as dust, pollen, microorganisms, etc.) exceeds the specified level, it may lead to product contamination, equipment failure, or distortion of experimental data, directly affecting product quality, production efficiency, and scientific research accuracy.

[0003] Traditionally, the environmental control of clean operation rooms relies on high-efficiency particulate air (HEPA) filters to remove particles in the air. At the same time, a positive pressure environment and air flow rate in the operation room are maintained through a fixed-set blower system to achieve the required cleanliness level. However, this static control strategy has some limitations. On the one hand, this static control strategy cannot immediately respond to changes in the actual cleanliness requirements in the operation room. On the other hand, as the HEPA filter is used for a longer time, it will gradually become blocked, resulting in an increase in pressure drop and affecting the filtration efficiency. The traditional method usually only replaces the filter at fixed time intervals and cannot adjust the air flow rate immediately according to the performance change of the filter, which may cause unnecessary energy consumption or premature failure of the filter.

[0004] Therefore, an optimized clean environment control system for a clean cold operation room is expected. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. The embodiments of this application provide a clean environment control system for a clean cold operation room. By collecting the air particulate matter concentration and the pressure data before and after the HEPA filter in the operation room in real time, and using time series analysis technology based on deep learning to perform time series modeling on the air particulate matter concentration data and the pressure drop data before and after the filter, it captures the changes in the clean environment in the operation room and the changes in the filter load state. Then, through time series correlation response analysis of the air particulate matter concentration and the filter pressure drop, it explores the potential correlation patterns between the air cleanliness in the operation room and the working state of the filter, so as to dynamically adjust the air volume of the blower. In this way, it can achieve adaptive control of the air cleanliness in the clean operation room, optimize the use efficiency of the HEPA filter, and reduce unnecessary energy consumption and filter replacement frequency.

[0006] Accordingly, according to an aspect of the present application, a clean environment control system for a clean cold operation room is provided, which includes:

[0007] A data acquisition module for acquiring the time series of air particulate matter concentration values, the time series of filter inlet pressure, and the time series of filter outlet pressure;

[0008] A filter pressure drop calculation module for performing time series alignment on the time series of the filter inlet pressure and the time series of the filter outlet pressure, and respectively calculating the difference between the filter inlet pressure and the filter outlet pressure at the same time step to obtain the time series of the filter pressure drop;

[0009] A time series encoding module for respectively extracting the local time series features of the time series of the air particulate matter concentration values and the time series of the filter pressure drop to obtain an air particulate matter concentration local time series correlation feature vector and a filter pressure drop local time series correlation feature vector;

[0010] A filtration effect response analysis module for performing cross-domain parameter core time series feature interaction response analysis on the air particulate matter concentration local time series correlation feature vector and the filter pressure drop local time series correlation feature vector to obtain a particulate matter concentration - filter pressure drop time series feature response interaction coding vector;

[0011] An air volume control module for determining whether to increase the air volume of the air supply fan based on the particulate matter concentration - filter pressure drop time series feature response interaction coding vector.

[0012] Compared with the prior art, the clean environment control system for a clean cold operation room provided by the present application captures the changes in the clean environment in the operation room and the changes in the filter load state by collecting the air particulate matter concentration in the operation room and the pressure data before and after the HEPA filter in real time, and using time series analysis technology based on deep learning to perform time series modeling on the air particulate matter concentration data and the pressure drop data before and after the filter. Furthermore, by performing time series correlation response analysis on the air particulate matter concentration and the filter pressure drop, the potential correlation pattern between the air cleanliness in the operation room and the working state of the filter is excavated, so as to dynamically adjust the air volume of the air supply fan. In this way, the adaptive control of the air cleanliness in the clean operation room can be realized, while optimizing the use efficiency of the HEPA filter and reducing unnecessary energy consumption and filter replacement frequency. Description of the Drawings

[0013] The above and other objects, features, and advantages of the present application will become more apparent by describing the embodiments of the present application in more detail with reference to the accompanying drawings. The drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation on the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0014] Figure 1 It is a block diagram of a clean environment control system for a clean cold operation room according to an embodiment of the present application.

[0015] Figure 2 It is a schematic diagram of data flow of a clean environment control system for a clean cold operation room according to an embodiment of the present application.

[0016] Figure 3 It is a block diagram of a filtration effect response analysis module in a clean environment control system for a clean cold operation room according to an embodiment of the present application.

[0017] Figure 4 It is a block diagram of a multi-granularity interaction response coding unit in a clean environment control system for a clean cold operation room according to an embodiment of the present application. Detailed implementation manners

[0018] As shown in the present application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "including" and "comprising" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0019] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0020] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in sequence. On the contrary, as needed, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0021] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described here.

[0022] In view of the technical problems described in the above background, the present application proposes a clean environment control system for a clean cold operation room. By collecting the air particulate matter concentration in the operation room and the pressure data before and after the HEPA filter in real time, and using the time series analysis technology based on deep learning to perform time series modeling on the air particulate matter concentration data and the pressure drop data before and after the filter, it captures the changes in the clean environment in the operation room and the changes in the filter load state. Furthermore, through the time series correlation response analysis of the air particulate matter concentration and the filter pressure drop, the potential correlation pattern between the air cleanliness in the operation room and the working state of the filter is excavated, so as to dynamically adjust the air volume of the air blower. In this way, the adaptive control of the air cleanliness in the clean operation room can be realized, while optimizing the use efficiency of the HEPA filter, reducing unnecessary energy consumption and the filter replacement frequency.

[0023] Figure 1 FIG. is a block diagram of a clean environment control system for a clean cold operation room according to an embodiment of the present application. Figure 2 FIG. is a schematic diagram of data flow of a clean environment control system for a clean cold operation room according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the clean environment control system 100 for the clean cold operation room includes: a data acquisition module 110 for acquiring the time series of the air particulate matter concentration value, the time series of the filter inlet pressure, and the time series of the filter outlet pressure; a filter pressure drop calculation module 120 for performing time series alignment on the time series of the filter inlet pressure and the time series of the filter outlet pressure, and respectively calculating the difference between the filter inlet pressure and the filter outlet pressure at the same time step to obtain the time series of the filter pressure drop; a time series encoding module 130 for respectively extracting the local time series features of the time series of the air particulate matter concentration value and the time series of the filter pressure drop to obtain the local time series correlation feature vector of the air particulate matter concentration and the local time series correlation feature vector of the filter pressure drop; a filtration effect response analysis module 140 for performing cross-domain parameter core time series feature interaction response analysis on the local time series correlation feature vector of the air particulate matter concentration and the local time series correlation feature vector of the filter pressure drop to obtain the particulate matter concentration-filter pressure drop time series feature response interaction coding vector; and an air volume control module 150 for determining whether to increase the air volume of the air blower based on the particulate matter concentration-filter pressure drop time series feature response interaction coding vector.

[0024] In the above-mentioned clean cold operation room clean environment control system, the data acquisition module 110 is used to acquire the time series of air particulate matter concentration values, the time series of filter inlet pressure, and the time series of filter outlet pressure. In a specific example of the present application, the time series of air particulate matter concentration values is collected by a particulate concentration sensor deployed in the operation room, and the time series of filter inlet pressure and the time series of filter outlet pressure are respectively collected by a first pressure sensor and a second pressure sensor installed before and after the HEPA filter. It should be understood that the clean environment of the operation room has a crucial impact on the safety of production activities and product quality. As a core indicator for measuring cleanliness, the change of air particulate matter concentration can intuitively reflect the clean state of the environment in the operation room. Therefore, by collecting air particulate matter concentration values in real time and arranging them as time series data, the present application can intuitively reflect the change trend of cleanliness in the operation room, providing accurate data support for subsequent operations such as adjusting the air volume of the air supply fan. At the same time, HEPA (High Efficiency Particulate Air Filter) is a key device for maintaining the clean environment of the operation room, and its filtering effect directly affects the air quality in the operation room. As the use time of the filter increases, impurities such as dust and particles will gradually accumulate on the surface and inside of the filter, resulting in an increase in the resistance of the filter, and thus affecting its filtering efficiency. By monitoring the pressure data at the inlet and outlet of the filter in real time, the present application can accurately understand the load state of the filter at different times, judge whether it is working properly and the change of the degree of blockage, so as to make appropriate adjustments to avoid premature failure of the filter or excessive energy consumption.

[0025] Specifically, the particulate concentration sensor is usually installed at key positions in the operation room and can monitor the change of particulate matter concentration in the air in real time. To ensure the accuracy and reliability of the data, the selection of the sensor is crucial. Sensors with high sensitivity and fast response time are preferred so that any fluctuations in particulate matter concentration can be captured in a timely manner. In addition, considering the differences in the spatial layout of different operation rooms, it may be necessary to install sensors at multiple positions to comprehensively cover the entire area, thereby obtaining more accurate data.

[0026] The first pressure sensor is installed before the HEPA filter and is used to measure the air pressure before entering the filter; while the second pressure sensor is installed after the filter and is responsible for recording the air pressure after filtration treatment. These two sets of data are extremely crucial for evaluating the working state of the filter. As the use time of the filter increases, dust and other impurities will gradually accumulate inside it, resulting in an increase in the pressure drop of the filter. By continuously monitoring the pressure difference between the front and rear ends of the filter, the degree of blockage of the filter and its filtering efficiency can be effectively evaluated. It should be noted that during the actual installation process, it is necessary to ensure that the position of the sensor is reasonably selected, avoiding direct contact with substances that may damage the sensor, while ensuring that it can accurately reflect the actual working condition of the filter.

[0027] To ensure that the collected data has high quality and accuracy, the calibration and maintenance of sensors cannot be ignored. Regularly calibrate the sensors to correct possible drift errors and ensure long-term stable data output. In addition, for operating rooms with long-term operations, a detailed maintenance plan is also required, including measures such as cleaning the sensor surface and checking whether the connection lines are intact, to prevent data anomalies caused by external factor interference. Only in this way can it be ensured that the time series of the air particulate matter concentration values, the inlet pressure and outlet pressure of the filter obtained from the clean cold operating room are both real and reliable, laying a solid foundation for further data analysis.

[0028] In practical applications, considering the changes in environmental conditions and the specific requirements of the operating room, corresponding adjustments may need to be made to the sensor configuration. For example, in some special cases, such as when manufacturing particularly sensitive products in the operating room, additional sensors may be required to improve the monitoring accuracy. Or in the case of a complex operating room structure and a large area, a distributed sensor network can be used to achieve more detailed spatial coverage, thereby ensuring that the air quality in every corner can be effectively monitored. In addition, for different production processes and cleanliness requirements, the sampling frequency and data transmission strategy of the sensors can also be customized to adapt to various complex industrial scenarios. In this way, not only can the flexibility and adaptability of the overall system be improved, but also the high-standard quality control requirements of the industry can be better met. In short, through the effective monitoring of the air particulate matter concentration values, the inlet pressure and outlet pressure of the filter, solid data support is provided for subsequent data analysis and intelligent control.

[0029] In the above-mentioned clean cold operation room clean environment control system, the filter pressure drop calculation module 120 is used to perform time series alignment on the time series of the filter inlet pressure and the time series of the filter outlet pressure, and calculate the difference between the filter inlet pressure and the filter outlet pressure at the same time step respectively to obtain the time series of the filter pressure drop. It should be understood that the pressure drop at the filter inlet and outlet directly reflects the resistance when the air flow passes through, and is an important index for evaluating the performance of the filter. The increase in the pressure drop means an increase in the degree of filter blockage, which will lead to a decrease in the filtration efficiency and an increase in the system energy consumption. Therefore, the present application further realizes the real-time monitoring of the change of the filter performance by calculating the filter pressure drop at each time point. Here, due to the influence of various factors such as the sensor response time and data transmission delay in the acquisition process of the filter inlet pressure and outlet pressure, the data points in the time series of the filter inlet pressure and the time series of the filter outlet pressure may not be strictly synchronized. Therefore, in order to accurately calculate the real pressure drop of the filter at each moment, first, based on the timestamp information of each data point, the time series of the filter inlet pressure and the time series of the filter outlet pressure are aligned in time series. Then, for each pair of time series-aligned inlet pressure and outlet pressure data, by calculating their difference, the time series of the filter pressure drop is obtained, so as to intuitively reflect the load situation of the filter at different time points and the change trend of its performance over time.

[0030] Specifically, first, it is necessary to preprocess the time series of the filter inlet pressure and the time series of the filter outlet pressure. Since there may be a slight time delay between the two sensors due to different physical positions or different signal transmission paths, it is a crucial step to ensure that the two sets of time series are compared on the same time basis. This usually involves precise matching of timestamps, and high-precision time synchronization technologies such as the Network Time Protocol (NTP) or more advanced Precision Time Protocol (PTP) are used to ensure time consistency among all devices. In this way, the time series of the filter inlet pressure and the filter outlet pressure can be adjusted to the same time axis, providing an accurate basis for subsequent calculations.

[0031] Next, calculate the difference between the filter inlet pressure and the outlet pressure at the same time step to obtain the time series of the filter pressure drop. This process involves extracting the filter inlet pressure value and the outlet pressure value from each corresponding time point and then performing a subtraction operation. Specifically, for each time step, subtract the corresponding outlet pressure from the filter inlet pressure, and the result represents the pressure drop of the filter at that moment. In actual operation, a multiple verification mechanism can be adopted to verify the correctness of each step. For example, by cross-comparing data in different time periods or introducing redundant calculation nodes, ensure that each calculation can be independently verified, thereby improving the reliability of the entire process.

[0032] During the execution of the above calculations, it is also necessary to consider possible data anomalies. Due to changes in the external environment or other unforeseen factors, abnormal fluctuations may occur in the pressure readings at some time steps. In the face of such a situation, it is particularly necessary to pre-develop a set of coping strategies. For example, set a reasonable threshold range. When it is detected that the pressure value in a certain period exceeds the normal fluctuation range, automatically trigger an alarm and mark the relevant data points for subsequent manual review. At the same time, adopt filtering algorithms such as moving average filtering or Kalman filtering to smooth out short-term noise interference and ensure that the obtained time series of the filter pressure drop is more stable and reliable. These measures not only help to improve the quality of the data but also lay a solid foundation for subsequent in-depth analysis.

[0033] In the above-mentioned clean cold operating room clean environment control system, the time series encoding module 130 is used to extract local time series features of the time series of the air particulate matter concentration value and the time series of the filter pressure drop, respectively, to obtain a local time series correlation feature vector of the air particulate matter concentration and a local time series correlation feature vector of the filter pressure drop. In a specific example of the present application, the time series encoding module 130 is used to perform local time series feature extraction based on one-dimensional convolutional coding on the time series of the air particulate matter concentration value and the time series of the filter pressure drop, to obtain the local time series correlation feature vector of the air particulate matter concentration and the local time series correlation feature vector of the filter pressure drop. In other words, in order to capture the temporal variation trends of the air particulate matter concentration and the filter pressure drop, the present application uses a one-dimensional convolutional neural network to extract temporal features from the time series of the air particulate matter concentration value and the time series of the filter pressure drop, respectively. Those skilled in the art will appreciate that a one-dimensional convolutional neural network has powerful local feature extraction capabilities. By performing a one-dimensional sliding convolution operation along the time dimension on the time series data, it can effectively capture local variation trends in the time series data. In the present application, by performing local time series feature extraction based on one-dimensional convolutional coding on the time series of the air particulate matter concentration value and the time series of the filter pressure drop, the local receptive field characteristics of the one-dimensional convolutional neural network can be utilized to mine the local dynamic change pattern of the air particulate matter concentration and the filter pressure drop in the time dimension, thereby obtaining the local time series correlation feature vector of the air particulate matter concentration and the local time series correlation feature vector of the filter pressure drop, providing effective feature input for subsequent blower air volume control decisions.

[0034] In the above-mentioned clean cold operation room clean environment control system, the filtration effect response analysis module 140 is used to perform cross-domain parameter core time-series feature interaction response analysis on the local time-series correlation feature vector of the air particulate matter concentration and the local time-series correlation feature vector of the filter pressure drop to obtain a particulate matter concentration-filter pressure drop time-series feature response interaction coding vector. It should be understood that there is a complex interaction relationship between the air particulate matter concentration and the filter pressure drop. For example, when the air particulate matter concentration in the operation room increases, it may mean that more dust and impurities have entered the operation room, which may cause an increase in the load of the HEPA filter, and thus increase its pressure drop. Conversely, when the pressure drop of the filter increases, it may mean that the filtration efficiency of the filter is decreasing, which may cause an increase in the air particulate matter concentration in the operation room. Therefore, in order to accurately capture the time-series interaction response pattern between the air particulate matter concentration and the filter pressure drop, the present application proposes a cross-domain parameter core time-series feature interaction response analysis method, which mines the core time-series information of the local time-series correlation feature vector of the air particulate matter concentration and the local time-series correlation feature vector of the filter pressure drop, and based on the core time-series information of the two, introduces a multi-level interaction response coding mechanism to achieve an accurate modeling of the time-series response relationship between the air particulate matter concentration and the filter pressure drop, so as to obtain a particulate matter concentration-filter pressure drop time-series feature response interaction coding vector. The particulate matter concentration-filter pressure drop time-series feature response interaction coding vector not only contains the time-series features of the air particulate matter concentration and the filter pressure drop respectively, but also integrates the time-series interaction response pattern features between the two, providing a more reliable decision-making basis for the subsequent air supply fan air volume control. Among them, Figure 3 is a block diagram of the filtration effect response analysis module in the clean cold operation room clean environment control system according to an embodiment of the present application. As Figure 3 shown, the filtration effect response analysis module 140 includes: a core information anchoring unit 141, which is used to extract the core information of the local time-series correlation feature vector of the air particulate matter concentration and the local time-series correlation feature vector of the filter pressure drop respectively to obtain an air particulate matter concentration time-series core information anchoring coding vector and a filter pressure drop time-series core information anchoring coding vector; a multi-granularity interaction response coding unit 142, which is used to perform multi-granularity interaction response coding on the air particulate matter concentration time-series core information anchoring coding vector and the filter pressure drop time-series core information anchoring coding vector to obtain the particulate matter concentration-filter pressure drop time-series feature response interaction coding vector.

[0035] Specifically, in a specific example of the present application, the core information anchoring unit 141 is used to: construct a temporal autocorrelation correlation matrix of the local temporal correlation feature vector of the air particulate matter concentration and the local temporal correlation feature vector of the filter pressure drop to obtain an air particulate matter concentration temporal autocorrelation correlation matrix and a filter pressure drop temporal autocorrelation correlation matrix, which is expressed by the formula:

[0036]

[0037]

[0038] Wherein, represents the transpose of the vector, represents the local temporal correlation feature vector of the air particulate matter concentration, represents the local temporal correlation feature vector of the filter pressure drop, represents the air particulate matter concentration temporal autocorrelation correlation matrix, represents the filter pressure drop temporal autocorrelation correlation matrix, is a linear mapping function.

[0039] That is, by constructing the temporal autocorrelation correlation matrix of the local temporal correlation feature vector of the air particulate matter concentration and the local temporal correlation feature vector of the filter pressure drop, the internal correlation structures of the local temporal correlation features of the air particulate matter concentration and the local temporal correlation features of the filter pressure drop are revealed respectively. In a specific implementation, the mutual correlation between the constituent feature parts in the vector is made explicit through vector outer product operation to generate an air particulate matter concentration temporal autocorrelation correlation matrix and a filter pressure drop temporal autocorrelation correlation matrix, so as to capture the internal semantic patterns of both while retaining the global temporal feature structure of the air particulate matter concentration and the filter pressure drop.

[0040] Specifically, in a specific example of the present application, the core information anchoring unit 141 is further used to: input the air particulate matter concentration temporal autocorrelation correlation matrix and the filter pressure drop temporal autocorrelation correlation matrix into a core information anchoring network based on autocorrelation decoupling respectively to obtain the air particulate matter concentration temporal core information anchoring coding vector and the filter pressure drop temporal core information anchoring coding vector, which is expressed by the formula:

[0041]

[0042]

[0043] Wherein, represents the core information anchoring network, represents the decoupling function, 、 、 and respectively represent the 1st, 2nd, th, and th row vectors in the autocorrelation correlation matrix of the time series of air particulate matter concentration, is the number of rows of the autocorrelation correlation matrix of the time series of air particulate matter concentration, , , and respectively represent the 1st, 2nd, th, and th row vectors in the autocorrelation correlation matrix of the time series of filter pressure drop, and respectively represent the time series feature weight matrix of air particulate matter concentration and the time series feature weight matrix of filter pressure drop, and respectively represent the time series bias vector of air particulate matter concentration and the time series bias vector of filter pressure drop, represents the conversion vector of the importance score of the local time series features of air particulate matter concentration, represents the conversion vector of the importance score of the local time series features of filter pressure drop, represents matrix multiplication, represents corresponding importance score factor of the local time series features of air particulate matter concentration, represents corresponding importance score factor of the local time series features of filter pressure drop, represents corresponding normalized importance score factor of the time series of air particulate matter concentration, represents corresponding normalized importance score factor of the time series of filter pressure drop, represents the sigmoid function, represents the core information anchoring coding vector of the time series of air particulate matter concentration, represents the core information anchoring coding vector of the time series of filter pressure drop.

[0044] Here, the present application uses a core information anchoring network based on autocorrelation decoupling to refine and purify the internal information of features of the time-series autocorrelation correlation matrix of air particulate matter concentration and the time-series autocorrelation correlation matrix of filter pressure drop. By performing feature decoupling, local feature importance evaluation, and feature weighted aggregation on the two, the feature parts highly correlated with the core semantics in the time-series autocorrelation correlation matrix of air particulate matter concentration and the time-series autocorrelation correlation matrix of filter pressure drop are highlighted, and at the same time, redundant information is denoised, so as to realize the time-series core feature anchoring of air particulate matter concentration and filter pressure drop, and obtain the time-series core information anchoring coding vector of air particulate matter concentration and the time-series core information anchoring coding vector of filter pressure drop.

[0045] Figure 4 It is a block diagram of a multi-granularity interaction response coding unit in a clean cold operation room clean environment control system according to an embodiment of the present application. As Figure 4 shown, the multi-granularity interaction response coding unit 142 includes: a feature granularity interaction response coding subunit 1421, configured to perform feature granularity response interaction coding on the time-series core information anchoring coding vector of the air particulate matter concentration and the time-series core information anchoring coding vector of the filter pressure drop to obtain a particulate matter concentration-filter pressure drop time-series feature granularity response interaction coding vector; a feature value granularity interaction response coding subunit 1422, configured to perform feature value granularity response interaction coding on the time-series core information anchoring coding vector of the air particulate matter concentration and the time-series core information anchoring coding vector of the filter pressure drop to obtain a particulate matter concentration-filter pressure drop time-series feature value granularity response interaction coding vector; and a fusion subunit 1423, configured to fuse the particulate matter concentration-filter pressure drop time-series feature granularity response interaction coding vector and the particulate matter concentration-filter pressure drop time-series feature value granularity response interaction coding vector to obtain the particulate matter concentration-filter pressure drop time-series feature response interaction coding vector.

[0046] More specifically, the feature granularity interaction response coding subunit 1421 is represented by the formula:

[0047]

[0048] where represents function, represents the particulate matter concentration-filter pressure drop time-series interaction feature weight matrix, represents the particulate matter concentration-filter pressure drop time-series interaction feature bias vector, represents the particulate matter concentration-filter pressure drop time-series feature granularity response interaction coding vector.

[0049] That is, after completing the core information anchoring, at the feature granularity level, by mining the response relationship between the air particulate matter concentration time series core information anchoring coding vector and the filter pressure drop time series core information anchoring coding vector on the time series feature unit, the time series feature coupling at the micro level is captured, and the particulate matter concentration-filter pressure drop time series feature granularity response interaction coding vector is obtained.

[0050] More specifically, the feature value granularity interactive response encoding subunit 1422 is expressed as follows:

[0051]

[0052] in, The particle size response interaction encoding vector representing the particle concentration-filter pressure drop time series characteristic value.

[0053] That is, at the eigenvalue granularity level, the present application performs element-by-element interactive response encoding on the air particulate matter concentration time series core information anchor coding vector and the filter pressure drop time series core information anchor coding vector to characterize the subtle correlation and mutual dependence between the two in terms of timing changes, and generates a particulate matter concentration-filter pressure drop time series eigenvalue granularity response interactive coding vector.

[0054] In particular, in a preferred example of the present application, the fusion subunit 1423 is configured to perform a bidirectional interactive characteristic distribution gradient constraint correction on the particle concentration-filter pressure drop time series characteristic particle size response interaction coding vector and the particle concentration-filter pressure drop time series characteristic value particle size response interaction coding vector to obtain an optimized particle concentration-filter pressure drop time series characteristic particle size response interaction coding vector and an optimized particle concentration-filter pressure drop time series characteristic value particle size response interaction coding vector, which are expressed as follows:

[0055]

[0056]

[0057] in, yes Middle The eigenvalues at the positions, yes Middle The eigenvalues at the positions, represents the cosine function, yes The corresponding optimized eigenvalue is, yes The corresponding optimized eigenvalue is, is the optimized particle concentration-filter pressure drop time series characteristic particle size response interaction encoding vector, It is an optimized particulate matter concentration - filter pressure drop time - series eigenvalue granularity response interaction coding vector.

[0058] Here, the present application takes into account that the difference between the modeling representations of feature granularity interaction and eigenvalue granularity interaction may cause unstable perturbations in the feature manifold interface of the unified coding vector that combines multi - granularity information expression. Therefore, preferably, the present application further uses the particulate matter concentration - filter pressure drop time - series eigenvalue granularity response interaction coding vector as an extensibility constraint representation. For the overall feature interaction distribution growth under the eigenvalue - by - eigenvalue diffusion process, based on the growth index representation under extensibility constraints, it models the interface shape perturbation deviation of the overall distribution of the particulate matter concentration - filter pressure drop time - series granularity response interaction coding vector interaction. That is, for the interaction interface gradient under each eigenvalue granularity as the perturbation contribution factor, it determines the growth index stabilization contribution of the gradient diffusion under eigenvalue interdependence to achieve the growth mode gradient correction dominated by perturbation stabilization, and obtains the optimized particulate matter concentration - filter pressure drop time - series granularity response interaction coding vector and the optimized particulate matter concentration - filter pressure drop time - series eigenvalue granularity response interaction coding vector.

[0059] More specifically, in a specific example of the present application, the fusion subunit 1423 is further configured to: perform cascade fusion on the optimized particulate matter concentration - filter pressure drop time - series granularity response interaction coding vector and the optimized particulate matter concentration - filter pressure drop time - series eigenvalue granularity response interaction coding vector to obtain the particulate matter concentration - filter pressure drop time - series feature response interaction coding vector, which is expressed by the formula:

[0060]

[0061] where, represents the particulate matter concentration - filter pressure drop time - series feature response interaction coding vector.

[0062] That is, cascade the optimized particulate matter concentration - filter pressure drop time - series granularity response interaction coding vector and the optimized particulate matter concentration - filter pressure drop time - series eigenvalue granularity response interaction coding vector to construct a unified coding vector that combines multi - granularity information expression, namely the particulate matter concentration - filter pressure drop time - series feature response interaction coding vector, so as to more completely describe the dynamic changes and mutual correlation patterns of particulate matter concentration and filter pressure drop.

[0063] In the clean environment control system of the above-mentioned clean cold operation room, the air volume control module 150 is used to determine whether to increase the air volume of the air supply fan based on the particulate matter concentration - filter pressure drop time series feature response interaction coding vector. In a specific example of the present application, the air volume control module 150 is used to: input the particulate matter concentration - filter pressure drop time series feature response interaction coding vector into the classifier-based control module to obtain a control instruction, and the control instruction is used to indicate whether to increase the air volume of the air supply fan. It should be understood that the adjustment of the air volume of the air supply fan needs to comprehensively consider the particulate matter concentration and the filter pressure drop. When the particulate matter concentration increases but the filter pressure drop is not too high, it indicates that the filtration efficiency of the filter still remains at a relatively high level. At this time, the air volume of the air supply fan should be appropriately increased to accelerate the air flow through the HEPA filter, increase the air circulation frequency, and improve the particle capture rate; when the particulate matter concentration increases and the filter pressure drop is also large, it may mean that the filter is already close to the saturation state. Simply increasing the air volume of the air supply fan may not effectively improve the air quality, but will instead increase the system energy consumption. At this time, the air volume of the air supply fan should be appropriately reduced to reduce the air flow through the blocked filter, avoid premature failure of the filter, and at the same time enable the standby HEPA or expand the system air duct load; when the particulate matter concentration is stable and the filter pressure drop is also maintained at a low level, it indicates that the current system is in a stable state and the air quality is good. At this time, the air volume of the air supply fan should be kept unchanged to maintain the stable operation of the system. In the technical solution of the present application, a classifier is used as the core of the control module. By performing feature pattern learning on the particulate matter concentration - filter pressure drop time series feature response interaction coding vector, the time series change law and interaction mode of the particulate matter concentration and the filter pressure drop are understood, the current air quality state and potential filtration system performance problems are identified, and thus the control instruction for the air volume of the air supply fan is intelligently output to guide the implementation of the air volume adjustment strategy of the air supply fan. In this way, the dynamic control and efficient management of the air quality in the operation room can be realized, effectively avoiding production interruption or product quality problems caused by the deterioration of the air quality, and at the same time, the service life of the HEPA filter can be extended, and the overall operation and maintenance cost of the system can be reduced.

[0064] More specifically, the air volume control module 150 is further configured to: perform a fully connected encoding on the particulate matter concentration - filter pressure drop time - series feature response interaction encoding vector using the fully connected layer of the control module to obtain a particulate matter concentration - filter pressure drop time - series feature response interaction fully connected encoding vector; input the particulate matter concentration - filter pressure drop time - series feature response interaction fully connected encoding vector into the Softmax classification function of the control module to obtain the probability values of the particulate matter concentration - filter pressure drop time - series feature response interaction encoding vector belonging to each classification label, where the classification labels include increasing the air volume of the air supply fan and not increasing the air volume of the air supply fan; determine the classification label corresponding to the maximum value among the probability values as the control instruction. Specifically, the fully connected layer can fully explore the deep - level information in the data, enhancing the expression ability and generalization performance of the model. The probability values output by the Softmax classification function provide an intuitive measure of the feasibility of different operation strategies, facilitating more scientific and reasonable decision - making.

[0065] Here, the local time - series correlation feature vector of the air particulate matter concentration and the local time - series correlation feature vector of the filter pressure drop respectively represent the time - series fluctuation mode features of the air particulate matter concentration and the time - series fluctuation mode features of the filter pressure drop. Considering that there may be noise or outliers in the source - domain data, which will cause the feature outliers to be misrecognized as core information during the feature interaction response analysis based on the core information anchoring, resulting in a fine - grained interaction response dynamic disorder offset in the particulate matter concentration - filter pressure drop time - series feature response interaction encoding vector, reducing the accuracy of the decoding result obtained by inputting into the control module based on the classifier.

[0066] In a preferred example, inputting the particulate matter concentration - filter pressure drop time - series feature response interaction encoding vector into the control module based on the classifier to obtain the control instruction includes:

[0067] Performing statistical analysis on the particulate matter concentration - filter pressure drop time - series feature response interaction encoding vector to obtain the mean value of the feature values and the standard deviation of the feature values ;

[0068] Based on the mean value of the feature values and the standard deviation of the feature values , constructing a particulate matter concentration - filter pressure drop time - series feature response interaction correlation weight threshold limit value, denoted as:

[0069]

[0070] where represents the length of the particulate matter concentration - filter pressure drop time - series feature response interaction encoding vector, Denote the threshold limit value of the interaction correlation weight of the particulate matter concentration-filter pressure drop time series feature response;

[0071] Based on the threshold limit value of the interaction correlation weight of the particulate matter concentration-filter pressure drop time series feature response, cooperate with the mean value of the feature values and the standard deviation of the feature values , perform cooperative balance benchmark mapping on the interaction coding vector of the particulate matter concentration-filter pressure drop time series feature response to obtain the first interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response and the second interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response, denoted as:

[0072]

[0073]

[0074] Wherein, denotes the interaction coding vector of the particulate matter concentration-filter pressure drop time series feature response, denotes dot product by position, denotes subtraction by position, denotes the first interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response, denotes the second interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response;

[0075] Based on the first interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response and the second interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response, perform dynamic conduction calibration on the interaction coding vector of the particulate matter concentration-filter pressure drop time series feature response to obtain the interaction coding global equilibrium correction coding vector of the particulate matter concentration-filter pressure drop time series feature response, denoted as:

[0076]

[0077] Wherein, denotes calculating the reciprocal of the feature value of each position of the second interaction coding cooperative balance benchmark vector of the particulate matter concentration-filter pressure drop time series feature response, denotes the weight hyperparameter, denotes the interaction coding global equilibrium correction coding vector of the particulate matter concentration-filter pressure drop time series feature response;

[0078] Based on the mean value of the feature values and the standard deviation of the feature values Perform equalization compensation on the interaction coding global equilibrium correction coding vector of the particulate matter concentration-filter pressure drop time series feature response to obtain an optimized particulate matter concentration-filter pressure drop time series feature response interaction coding vector, expressed as:

[0079]

[0080] Wherein, Indicates addition by position, Indicates the optimized particulate matter concentration-filter pressure drop time series feature response interaction coding vector.

[0081] Input the optimized particulate matter concentration-filter pressure drop time series feature response interaction coding vector into the classifier-based control module to obtain the control instruction.

[0082] That is, by using the association weight threshold limit based on the particulate matter concentration-filter pressure drop time series feature response interaction coding vector as the collaborative balance benchmark mapping, calibrate the dynamic conduction of the set features of the particulate matter concentration-filter pressure drop time series feature response interaction coding vector, and use the aggregation feature response association based on the particulate matter concentration-filter pressure drop time series feature response interaction coding vector as the feature domain multi-level equalization compensation to achieve the global consistency equalization characterization of the particulate matter concentration-filter pressure drop time series feature response interaction coding vector, and form a fair collaborative architecture under the feature distribution framework of the particulate matter concentration-filter pressure drop time series feature response interaction coding vector. In this way, improve the accuracy of the control instruction obtained by inputting it into the classifier-based control module.

[0083] Specifically, after receiving the instruction, the controller will send a signal to the air blower, asking to adjust its working parameters to increase the air volume. After obtaining the instruction to increase the air volume of the air blower, the control module will send a signal to the variable frequency drive (VFD) of the air blower, notifying it to increase the output frequency, thereby increasing the rotational speed of the air blower. By adjusting the speed of the air blower, the air flow can be effectively increased to ensure that more fresh air is introduced into the clean cold operation room. At the same time, in order to balance the pressure difference between the newly introduced air and the existing air in the room, the control system also needs to synchronously adjust the opening degrees of the return air outlet and the exhaust air outlet. Appropriately increasing the opening degree of the return air outlet helps to quickly discharge the polluted air in the room, while adjusting the exhaust air outlet can effectively prevent the untreated external air from flowing back into the operation room, ensuring the positive pressure state of the entire system.

[0084] In addition, during the process of increasing the air volume of the air blower, the monitoring sensors keep working, providing real-time feedback on the air quality and air flow conditions in the operation room. These sensors not only monitor the changes in particulate matter concentration but also detect other key parameters such as temperature and humidity to ensure that all conditions meet the set standards. If any deviation from the predetermined value is detected, the control system can make timely fine adjustments to avoid energy waste caused by excessive ventilation or a decline in air quality due to insufficient ventilation. For example, if the particulate matter concentration does not decrease as expected, the control system may further increase the air volume of the air blower; conversely, if the temperature or humidity exceeds the specified range, the air volume needs to be reduced accordingly or auxiliary adjustment equipment needs to be activated.

[0085] When the system determines that there is no need to increase the air volume of the air blower, it means that the current air quality already meets the requirements or only minor adjustments are needed to reach the ideal level. At this time, the control module keeps the existing air blower speed unchanged and maintains good air quality by finely regulating other environmental parameters. First of all, the control system may slightly adjust the opening degrees of the return air vent and the exhaust vent to optimize the internal air circulation path and promote the effective removal of pollutants. This strategy can enhance the air purification effect in local areas without significantly changing the total air volume, especially suitable for scenarios with high air quality requirements for specific locations.

[0086] In addition to adjusting the opening degrees of the ventilation vents, the built-in air purification device can also be used to assist in improving air quality. For example, activating the ultraviolet germicidal lamp or the ion generator, these devices can effectively kill microorganisms or remove harmful substances in the air without affecting the overall air volume. In addition, according to actual needs, the humidifier or dehumidifier can be enabled to precisely control the humidity level in the operation room to ensure that the environmental conditions always meet the requirements of the production process. In this way, both the stable control of air quality and the avoidance of unnecessary energy consumption are achieved.

[0087] While performing the above operations, the monitoring system continuously collects and analyzes various data, including particulate matter concentration, temperature and humidity changes, etc. Once any abnormal situation is detected, such as a sudden increase in particulate matter concentration or temperature and humidity fluctuations exceeding the allowable range, the control system will automatically trigger the warning mechanism to remind relevant personnel to pay attention and take corresponding measures. In this case, although it is not necessary to significantly adjust the air volume of the air blower, it may be necessary to temporarily activate the standby purification equipment or temporarily adjust some process flows to prevent potential problems from expanding.

[0088] In summary, the clean environment control system for a clean cold operation room based on the embodiments of the present application is elucidated. It captures the changes in the clean environment within the operation room and the changes in the filter load status by collecting the air particulate matter concentration in the operation room and the pressure data before and after the HEPA filter in real time, and performing time series modeling on the air particulate matter concentration data and the pressure drop data before and after the filter using time series analysis technology based on deep learning. Furthermore, by performing time series correlation response analysis on the air particulate matter concentration and the filter pressure drop, potential correlation patterns between the air cleanliness in the operation room and the working status of the filter are excavated, thereby dynamically adjusting the air volume of the air supply fan. In this way, adaptive control of the air cleanliness in the clean operation room can be achieved, while optimizing the usage efficiency of the HEPA filter, reducing unnecessary energy consumption and the filter replacement frequency.

[0089] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present invention are merely examples and not limitations. It cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present invention. Additionally, the specific details of the above embodiments are for illustrative and facilitating understanding purposes only, not limitations. The above details do not limit the present invention to necessarily adopt the above specific details for implementation.

[0090] In the above embodiments, the descriptions of each embodiment have their own focuses. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the unit division is only a logical function division, and there can be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0091] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, in any aspect, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be encompassed by the present invention. Any associated drawing reference signs in the claims should not be regarded as limiting the claimed rights.

[0092] Finally, it should be noted that the above description has been given for purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A clean environment control system for a clean cold operation room, characterized in that, Including: A data acquisition module, configured to acquire a time series of air particulate matter concentration values, a time series of filter inlet pressures, and a time series of filter outlet pressures; A filter pressure drop calculation module, configured to perform time series alignment on the time series of the filter inlet pressure and the time series of the filter outlet pressure, and calculate the difference between the filter inlet pressure and the filter outlet pressure at the same time step respectively to obtain a time series of filter pressure drops; A time series encoding module, configured to extract local time series features of the time series of the air particulate matter concentration values and the time series of the filter pressure drops respectively to obtain an air particulate matter concentration local time series correlation feature vector and a filter pressure drop local time series correlation feature vector; A filtration effect response analysis module, configured to perform cross-domain parameter core time series feature interaction response analysis on the air particulate matter concentration local time series correlation feature vector and the filter pressure drop local time series correlation feature vector to obtain a particulate matter concentration - filter pressure drop time series feature response interaction coding vector; An air volume control module, configured to determine whether to increase the air volume of the air blower based on the particulate matter concentration - filter pressure drop time series feature response interaction coding vector.

2. The clean environment control system for a clean cold operation room according to claim 1, wherein, The time series of the air particulate matter concentration values is collected by a particulate matter concentration sensor deployed in the operation room, and the time series of the filter inlet pressure and the time series of the filter outlet pressure are collected by a first pressure sensor and a second pressure sensor installed before and after the HEPA filter respectively.

3. The clean environment control system for the clean cold operation room according to claim 2, characterized in that, The time series encoding module is configured to: Perform local time series feature extraction based on one-dimensional convolutional encoding on the time series of the air particulate matter concentration values and the time series of the filter pressure drops to obtain the air particulate matter concentration local time series correlation feature vector and the filter pressure drop local time series correlation feature vector.

4. The clean cold operation room clean environment control system according to claim 3, characterized in that, The filtration effect response analysis module includes: A core information anchoring unit, configured to extract core information of the air particulate matter concentration local time series correlation feature vector and the filter pressure drop local time series correlation feature vector respectively to obtain an air particulate matter concentration time series core information anchoring coding vector and a filter pressure drop time series core information anchoring coding vector; A multi-granularity interaction response coding unit, configured to perform multi-granularity interaction response coding on the air particulate matter concentration time series core information anchoring coding vector and the filter pressure drop time series core information anchoring coding vector to obtain the particulate matter concentration - filter pressure drop time series feature response interaction coding vector.

5. The clean environment control system for a clean cold operation room according to claim 4, characterized in that, The core information anchoring unit is configured to: Construct a time series autocorrelation correlation matrix of the air particulate matter concentration local time series correlation feature vector and the filter pressure drop local time series correlation feature vector to obtain an air particulate matter concentration time series autocorrelation correlation matrix and a filter pressure drop time series autocorrelation correlation matrix; Input the air particulate matter concentration time series autocorrelation correlation matrix and the filter pressure drop time series autocorrelation correlation matrix into a core information anchoring network based on autocorrelation decoupling respectively to obtain the air particulate matter concentration time series core information anchoring coding vector and the filter pressure drop time series core information anchoring coding vector.

6. The clean environment control system for the clean cold operation room according to claim 5, characterized in that, The multi-granularity interaction response encoding unit includes: A feature granularity interaction response encoding subunit, configured to perform feature granularity response interaction encoding on the air particulate matter concentration time series core information anchored encoding vector and the filter pressure drop time series core information anchored encoding vector to obtain a particulate matter concentration - filter pressure drop time series feature granularity response interaction encoding vector; An eigenvalue granularity interaction response encoding subunit, configured to perform eigenvalue granularity response interaction encoding on the air particulate matter concentration time series core information anchored encoding vector and the filter pressure drop time series core information anchored encoding vector to obtain a particulate matter concentration - filter pressure drop time series eigenvalue granularity response interaction encoding vector; A fusion subunit, configured to fuse the particulate matter concentration - filter pressure drop time series feature granularity response interaction encoding vector and the particulate matter concentration - filter pressure drop time series eigenvalue granularity response interaction encoding vector to obtain the particulate matter concentration - filter pressure drop time series feature response interaction encoding vector.

7. The clean environment control system for a clean cold operation room according to claim 6, wherein, The fusion subunit is configured to: Perform feature distribution gradient constraint correction based on bidirectional interaction on the particulate matter concentration - filter pressure drop time series feature granularity response interaction encoding vector and the particulate matter concentration - filter pressure drop time series eigenvalue granularity response interaction encoding vector to obtain an optimized particulate matter concentration - filter pressure drop time series feature granularity response interaction encoding vector and an optimized particulate matter concentration - filter pressure drop time series eigenvalue granularity response interaction encoding vector; Perform cascade fusion on the optimized particulate matter concentration - filter pressure drop time series feature granularity response interaction encoding vector and the optimized particulate matter concentration - filter pressure drop time series eigenvalue granularity response interaction encoding vector to obtain the particulate matter concentration - filter pressure drop time series feature response interaction encoding vector.

8. The clean environment control system for a clean cold operation room according to claim 7, wherein The air volume control module is configured to: Input the particulate matter concentration - filter pressure drop time series feature response interaction encoding vector into a classifier-based control module to obtain a control instruction, where the control instruction is used to indicate whether to increase the air volume of the air blower.

9. The clean cold operation room clean environment control system according to claim 8, wherein The air volume control module is configured to: Perform fully connected encoding on the particulate matter concentration - filter pressure drop time series feature response interaction encoding vector using the fully connected layer of the control module to obtain a particulate matter concentration - filter pressure drop time series feature response interaction fully connected encoding vector; Input the particulate matter concentration - filter pressure drop time series feature response interaction fully connected encoding vector into the Softmax classification function of the control module to obtain the probability values of the particulate matter concentration - filter pressure drop time series feature response interaction encoding vector belonging to each classification label, where the classification labels include increasing the air volume of the air blower and not increasing the air volume of the air blower; Determine the classification label corresponding to the maximum value among the probability values as the control instruction.

Images