FFU group control linkage regulation system based on multi-parameter fusion and self-learning mechanism

Abstract

The application discloses a FFU group control linkage regulation system based on a multi-parameter fusion and self-learning mechanism, which comprises a group control cabinet main body, a parameter input module, a data acquisition module, an AI intelligent control module and a FFU linkage module. The parameter input module and the data acquisition module respectively acquire static and dynamic data. The AI intelligent control module is internally provided with a multi-parameter fusion unit, which is used for correlating and mapping static and dynamic parameters, constructing a digital twin model and calculating a theoretical basis instruction; a space collaborative regulation unit is used for dynamically dividing a collaborative regulation group according to a cleanliness distribution gradient to generate a differentiated instruction; and a device self-learning compensation unit is used for constructing a personalized compensation model based on an aging trend of each FFU to correct the theoretical basis instruction. The FFU linkage module outputs a final regulation value to realize single or grouped independent driving control. Through multi-parameter fusion, space collaboration and self-learning compensation, the application can significantly improve the cleanliness regulation accuracy and uniformity, eliminate local blind areas, reduce energy consumption and prolong the service life of the equipment.

Description

Technical Field

[0001] This invention relates to the field of wind turbine control technology, and in particular to an FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism. Background Technology

[0002] Fan filter units (FFUs) are core equipment for maintaining the cleanliness of cleanrooms, laboratories, and lithium battery and semiconductor manufacturing environments. Currently, FFU group control technology mainly suffers from the following two types of solutions and their inherent limitations:

[0003] First, operators need to adjust the speed of each FFU individually, or can only achieve unified start-stop and constant-speed operation of all FFUs. This method cannot be differentiated according to the actual size of the clean space, the differences in FFU layout, and the real-time cleanliness requirements of different areas. When there are many FFUs or the space is large, it is easy to cause uneven cleanliness distribution, resulting in local over-cleanliness or excessive cleanliness, and serious energy waste.

[0004] Secondly, while some systems use sensors to collect environmental parameters for closed-loop control, their control models are fixed and lack the ability to comprehensively analyze multi-dimensional input parameters. For example, relying solely on real-time sensor data for control leads to control strategies that are not adapted to the actual scenario, requiring repeated manual calibration. Furthermore, after long-term operation, filter clogging and motor aging cause the response curve of the actual airflow as a function of speed to drift. Existing systems cannot track the individual performance trends of each FFU, causing the initially set control commands to gradually become ineffective. This necessitates periodic manual recalibration, resulting in a large maintenance workload and a decrease in control accuracy over time. Summary of the Invention

[0005] To address at least one of the aforementioned technical problems, this invention provides an FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanisms.

[0006] In a first aspect, the present invention provides an FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism, the system comprising:

[0007] The main body of the group control cabinet;

[0008] The parameter input module is used to obtain static configuration parameters input by the user. The static configuration parameters include at least FFU performance parameters, clean space geometric parameters, FFU layout and quantity parameters, and target cleanliness level.

[0009] The data acquisition module is used to acquire dynamic operating parameters in real time. The dynamic operating parameters include at least the cleanliness data of multiple points in the space, the real-time operating parameters and status data of each FFU.

[0010] The AI ​​intelligent control module includes:

[0011] The multi-parameter fusion unit is used to map static configuration parameters to dynamic operating parameters, build an initial digital twin model of the clean space, and calculate the theoretical control commands for each FFU.

[0012] The spatial coordinated control unit is used to identify the cleanliness distribution gradient based on static configuration parameters and dynamic operating parameters, dynamically divide all FFUs into multiple coordinated control groups based on the identification results, and generate differentiated control instructions for different coordinated control groups.

[0013] The equipment self-learning compensation unit is used to build a personalized aging compensation model for each FFU based on the running time, filter resistance change curve and speed-airflow response deviation of each FFU throughout its entire life cycle. Based on the aging compensation model, the theoretical control commands are corrected to generate individual compensation commands for each FFU.

[0014] The FFU linkage module is used to receive differentiated control commands and individual compensation commands, determine the coordinated control group to which each FFU belongs based on the differentiated control commands, correct the benchmark control value of each FFU to which the coordinated control group belongs on a unit-by-unit basis based on the individual compensation commands, and output the final control command to independently drive and control each FFU or each coordinated control group.

[0015] Secondly, the present invention also provides an FFU group control and linkage regulation method based on multi-parameter fusion and self-learning mechanism, applied to the system as described in any one of the first aspects, comprising the following steps:

[0016] Obtain static configuration parameters input by the user and collect dynamic runtime parameters in real time;

[0017] By associating and mapping static configuration parameters with dynamic operating parameters, an initial digital twin model of the clean space is constructed, and the theoretical basic control commands for each FFU are calculated.

[0018] Based on the spatial cleanliness distribution gradient, all FFUs are dynamically divided into multiple collaborative control groups, generating differentiated group control instructions.

[0019] Based on the runtime, filter resistance changes, and response deviation of each FFU, its personalized aging compensation model is invoked to modify the basic control commands and group control commands for each unit, generating the final control commands.

[0020] Based on the final control instructions, each FFU or each coordinated control group is driven and controlled independently.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0022] 1) By using a multi-parameter fusion unit to map static configuration parameters (spatial geometry, FFU performance, layout) with dynamic operating parameters, a digital twin model of the clean space is constructed. The optimal control command is then iteratively solved based on the contribution weight of each FFU to the cleanliness of different locations within the space. Compared to existing coarse-grained control methods, this invention ensures that the output of each FFU precisely matches the actual needs, significantly improving the cleanliness compliance rate and uniformity, and stably meeting the requirements of different cleanliness levels from Class 100 to Class 10,000.

[0023] 2) The system generates a real-time cleanliness distribution heat map through a spatial collaborative control unit, automatically identifying compliant and non-compliant areas. It dynamically groups boundary FFUs and FFUs within non-compliant areas into the same collaborative control group, guiding clean airflow towards the weaker areas by increasing the rotation speed of the boundary FFUs. This mechanism endows the system with proactive "airflow scheduling" capabilities, effectively solving the problem of localized cleanliness deterioration and lag response in traditional group control systems, and achieving a collaborative balance of cleanliness across the entire space.

[0024] 3) Through the equipment self-learning compensation unit, the system can record the drift of the speed-airflow characteristic curve and the trend of filter resistance change of each FFU throughout its entire life cycle, and build a personalized aging compensation model (such as speed compensation coefficient). This model can automatically correct the airflow deviation caused by filter clogging and motor aging, so that old FFUs can still output the expected airflow, avoiding frequent replacement or manual calibration due to equipment performance degradation, extending the effective service life of the equipment, and reducing operation and maintenance costs.

[0025] 4) Because the system achieves precise output from each FFU on demand (based on the optimal solution of contribution weight) and airflow guidance based on spatial coordination (avoiding blindly increasing the speed of all FFUs), it avoids the "over-cleanliness" or "ineffective high-load operation" common in traditional control. The AI ​​intelligent control module automatically optimizes the total energy consumption while ensuring cleanliness, significantly reducing the overall energy consumption of the FFU group, which meets the requirements of green industrial production.

[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the background art, the accompanying drawings used in the embodiments of the present invention or the background art will be described below.

[0028] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the specification, serve to illustrate the technical solutions of this disclosure.

[0029] Figure 1A schematic diagram of the structure of an FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism provided in an embodiment of the present invention;

[0030] Figure 2 Provided for embodiments of the present invention Figure 1 A schematic diagram of the sub-modules of the AI ​​intelligent control module;

[0031] Figure 3 An overall structural block diagram of an FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism provided in an embodiment of the present invention;

[0032] Figure 4 This is a block diagram of the internal structure of the AI ​​intelligent control module provided in an embodiment of the present invention;

[0033] Figure 5 A flowchart illustrating an FFU group control and linkage regulation method based on multi-parameter fusion and self-learning mechanism provided for an embodiment of the present invention;

[0034] Figure 6 A system operation flowchart provided for embodiments of the present invention;

[0035] Figure 7 This is a schematic diagram of the main structure of the group control cabinet provided in an embodiment of the present invention;

[0036] Figure 8 This is a schematic diagram of the grouping of FFU linkage control provided in an embodiment of the present invention. Detailed Implementation

[0037] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0039] Existing FFU group control technologies mainly employ independent adjustment of individual units or unified start-stop control, making it difficult to coordinate and regulate based on space size, FFU layout, and real-time cleanliness differences. This easily leads to uneven cleanliness and energy waste. While feedback systems based on fixed logic incorporate sensor data, they fail to deeply integrate user-input static parameters with dynamic operating data, lacking the ability to adapt to performance drift caused by long-term equipment aging. Control accuracy decreases over time, resulting in high maintenance costs. Therefore, this invention provides an FFU group control and linkage system. By mapping static configuration parameters with real-time dynamic parameters, a characterization model of the clean space is constructed to calculate the theoretical control benchmark. Based on the spatial cleanliness distribution gradient, a collaborative relationship between FFUs is dynamically established to compensate for weak areas. Simultaneously, the system tracks the performance change trend of each FFU throughout its entire lifecycle and automatically corrects control commands, thereby achieving precise, energy-saving, and long-term stable group control. The invention will be further described below with reference to specific embodiments.

[0040] Please see Figure 1 , Figure 1 This is a schematic diagram of the structure of an FFU group control and linkage system based on multi-parameter fusion and self-learning mechanism, provided as an embodiment of the present invention. Figure 1 As shown, the system includes:

[0041] Group control cabinet main body 10;

[0042] The parameter input module 20 is used to obtain static configuration parameters input by the user. The static configuration parameters include at least FFU performance parameters, clean space geometric parameters, FFU layout and quantity parameters, and target cleanliness level.

[0043] Data acquisition module 30 is used to acquire dynamic operating parameters in real time. The dynamic operating parameters include at least the cleanliness data of multiple points in the space, the real-time operating parameters and status data of each FFU.

[0044] AI intelligent control module 40, including:

[0045] The multi-parameter fusion unit 401 is used to associate and map static configuration parameters with dynamic operating parameters to build an initial digital twin model of the clean space in order to calculate the theoretical basic control commands for each FFU.

[0046] The spatial coordinated control unit 402 is used to identify the cleanliness distribution gradient based on static configuration parameters and dynamic operating parameters, dynamically divide all FFUs into multiple coordinated control groups based on the identification results, and generate differentiated control instructions for different coordinated control groups.

[0047] The equipment self-learning compensation unit 403 is used to construct a personalized aging compensation model for each FFU based on the running time, filter resistance change curve and speed-airflow response deviation of each FFU throughout its entire life cycle. Based on the aging compensation model, the theoretical control commands are corrected to generate individual compensation commands for a single FFU.

[0048] The FFU linkage module 50 is used to receive differentiated control commands and individual compensation commands, determine the coordinated control group to which each FFU belongs based on the differentiated control commands, correct the benchmark control value of each FFU to which the coordinated control group belongs on a unit-by-unit basis based on the individual compensation commands, and output the final control command to independently drive and control each FFU or each coordinated control group.

[0049] In this embodiment, before the system is put into operation, the user inputs static configuration parameters through the parameter input module. These parameters include: the rated performance parameters of the FFU (such as rated air volume, rated power, initial filtration resistance, and factory-calibrated speed-air volume characteristic curve), the geometric dimensions and layout of the clean space, the installation location coordinates and quantity distribution grid of the FFUs, and the target cleanliness level that the space needs to achieve (corresponding to the upper limit of the concentration of particles of a specific size). After the input is completed, the parameter input module 20 packages these data and sends them to the AI ​​intelligent control module as a benchmark for subsequent modeling and control.

[0050] Data acquisition module 30 continuously acquires two types of dynamic operating parameters at a set sampling period:

[0051] Space cleanliness distribution data: The particle concentration values ​​at each measuring point are acquired in real time by using multiple fixed or mobile particle concentration sensors arranged in the clean space (recorded separately according to particle size channel).

[0052] Status parameters for each FFU: including but not limited to current speed, real-time air volume, motor current, filter resistance, cumulative running time, and the most recent speed-air volume response deviation record.

[0053] The multi-parameter fusion unit 401 associates and maps the received static configuration parameters with dynamic operating parameters to construct an initial digital twin model of the clean space. Specifically, this model integrates the following relationships:

[0054] Based on the FFU layout and spatial geometry parameters in the static parameters, a simplified simulation model of airflow organization is established, and the theoretical influence area of ​​each FFU and the coupling coefficient between adjacent FFUs are defined.

[0055] Based on the real-time cleanliness of each measuring point in the dynamic parameters, the cleanliness distribution gradient field in the space is calculated in reverse.

[0056] Using the target cleanliness level as a constraint and combining it with the airflow coupling coefficient matrix, the theoretical basic control command for each FFU is calculated by solving the optimization problem. That is, the theoretical speed or theoretical airflow set value required to achieve the overall cleanliness target without considering individual aging differences.

[0057] The spatial coordinated control unit 402 receives the cleanliness distribution gradient field output by the multi-parameter fusion unit and the real-time concentration data of each measuring point. Specifically:

[0058] Cleanliness gradient identification: Spatial clustering algorithms (such as DBSCAN or density peak-based methods) are used to identify areas that meet cleanliness standards, critical areas, and areas that exceed standards. Simultaneously, the direction and magnitude of the concentration gradient between adjacent areas are calculated to determine possible pathways for pollutant diffusion.

[0059] Dynamic Coordination and Control Group Division: Based on the gradient identification results, all FFUs are divided into multiple coordination and control groups. FFUs within each group are spatially adjacent, and the cleanliness change trends within their affected areas are consistent. The grouping rules are as follows:

[0060] FFUs located in areas with excessive cleanliness levels are classified as "enhanced group" and require increased air supply volume.

[0061] FFUs located in areas with acceptable and stable cleanliness levels are classified as "maintenance groups" to maintain current output or slightly reduce airflow to save energy.

[0062] FFUs located in the transition zone between the excessive and compliant areas are classified as "buffer groups" and are subject to moderate control commands to prevent new cleanliness imbalances caused by excessive local control.

[0063] Generate differentiated control commands: Generate a baseline control value (e.g., a set of shared airflow correction coefficients) for each coordinated control group. The baseline control values ​​differ between different groups. This baseline value is calculated based on the average or weighted average of the theoretical control commands of each FFU within the group, and is offset according to the control strategy (enhancement / buffering / maintenance) to which the group belongs.

[0064] The equipment self-learning compensation unit 403 establishes an independent lifecycle aging profile for each FFU, including:

[0065] Feature extraction: For each FFU, record its cumulative running time since commissioning, the operating cycle after each filter replacement, the filter resistance change curve over time, and the deviation sequence between the actual air volume response under different speed commands and the factory calibration curve (i.e., speed-air volume response deviation).

[0066] Model Construction: Based on the historical data mentioned above, a lightweight regression model (such as multinomial regression or recursive least squares algorithm with a forgetting factor) is used to fit the aging trend of each FFU online. This model can predict the ratio of the decrease in actual air volume to theoretical air volume due to factors such as filter dust accumulation and motor bearing wear under the current operating conditions, as well as the impact of increased filter resistance on the fan efficiency.

[0067] Individual Compensation Command Generation: For the theoretical control command of this FFU output by the multi-parameter fusion unit, the aging compensation model calculates the additional speed compensation or airflow compensation coefficient required based on the current operating time and recent resistance change rate. This compensation amount is not globally uniform, but is calculated independently for each FFU, ultimately generating an individual compensation command.

[0068] The FFU linkage module 50 receives differentiated control commands (baseline control values ​​for each group) from the spatial coordinated control unit and individual compensation commands (personalized correction amounts for each FFU) from the equipment self-learning compensation unit. Its fusion logic is as follows:

[0069] Based on the differentiated control commands, each FFU is assigned to a coordinated control group, thus obtaining the baseline control value for that group. On top of this baseline value, the individual compensation command for the corresponding FFU is superimposed (e.g., an FFU requiring an additional +3% speed compensation due to aging). The superposition method can be linear superposition or nonlinear fusion weighted by confidence level. The final control command (target speed or target airflow) is output and sent to the inverter driver or intelligent control terminal of each FFU via fieldbus.

[0070] During system operation, the data acquisition module continuously provides feedback on changes in cleanliness after adjustments and the actual operating parameters of each FFU. The AI ​​intelligent control module periodically (e.g., once per hour or after each adjustment action) re-executes multi-parameter fusion and spatial collaborative partitioning, and updates the aging compensation model for each FFU. When the filtration resistance of a certain FFU exceeds the set threshold or the speed-airflow response deviation continues to increase, the system automatically issues a filter replacement or maintenance warning, and resets the aging model of that FFU after replacement, restarting the learning process.

[0071] Through the above process, this embodiment realizes intelligent linkage control across the entire chain, from global cleanliness targets to group collaboration and individual unit personalized compensation. It can effectively reduce overall energy consumption and extend the overall service life of the FFU cluster while ensuring that cleanliness standards are met.

[0072] In one embodiment, the multi-parameter fusion unit 401 is further configured to:

[0073] Based on the geometric parameters of the clean space and the layout and number of FFUs, a spatial airflow organization influence matrix is ​​established. The matrix is ​​used to characterize the contribution weight of each FFU to the cleanliness of different locations in the space. The contribution weight is determined according to the physical distance and the obstacle obstruction factor.

[0074] Based on the difference between the target cleanliness level and the cleanliness data of multiple points in the current space, and combined with the contribution weight, the theoretical rotation speed and air volume of each FFU are solved in reverse iteration to minimize the difference and optimize the total energy consumption of all FFUs.

[0075] Specifically, establish a spatial airflow organization influence matrix. , where matrix elements Characterizing the first Taiwan FFU to the first The contribution weight of cleanliness at each spatial location is calculated using the following formula:

[0076]

[0077] In the formula, For the first Taiwan FFU, For the first Taiwan FFU wind center and the first Euclidean distance between spatial locations The output efficiency coefficient of the FFU ranges from 0.6 to 0.95, and is determined based on the FFU model and filter resistance characteristics. The airflow attenuation factor reflects the degree of cleanliness attenuation caused by factors such as obstacles, eddies, temperature, and humidity during airflow propagation. Its value ranges from 0.1 to 0.5 and is calibrated through on-site measurement or CFD simulation.

[0078] The above formula takes into account the FFU output volume, propagation distance, geometric attenuation, and the attenuation effect of the spatial environment on the airflow, and can reasonably reflect the actual contribution of each FFU to the cleanliness of the remote location.

[0079] Furthermore, based on the target cleanliness level required for the clean space (such as Class 5 or Class 6 in ISO 14644-1), the target particle concentration value at each detection point is determined. By deploying multiple cleanliness sensors throughout the space, the particle concentration at each detection point is collected in real time. Calculate the difference between the two. :

[0080]

[0081] In the formula, if A positive value indicates that the current cleanliness level is lower than the target value, and the air supply to the corresponding area needs to be increased; a negative value indicates that the cleanliness level has exceeded the target, and the air supply can be appropriately reduced to save energy.

[0082] Furthermore, based on the contribution weight matrix The difference in cleanliness at each testing point Construct the following optimization problem:

[0083] Find the theoretical speed adjustment (or air volume adjustment) for each FFU to minimize the weighted space cleanliness deviation while minimizing the total energy consumption of all FFUs.

[0084] The objective function can be expressed as:

[0085]

[0086] In the formula, : No. Airflow adjustment of the FFU (Fan Filter Unit);

[0087] : No. Taiwan FFU airflow adjustment The energy consumption function is typically a quadratic or linear model.

[0088] Energy consumption weighting coefficient, used to balance cleanliness deviations and energy-saving targets.

[0089] The solution method employs a reverse iterative optimization algorithm (such as gradient descent, least squares, or genetic algorithm) to iteratively calculate the optimal airflow adjustment under the following constraints: the FFU speed / airflow must not exceed its physical upper and lower limits; the rate of change of speed of each FFU does not exceed the safety threshold; the theoretical speed and airflow values ​​of each FFU obtained from the solution are converted into control signals and sent to the corresponding FFU controller to achieve dynamic and differentiated airflow organization and adjustment in the clean space.

[0090] By implementing this embodiment, the operating status of each FFU can be intelligently allocated according to the actual contamination distribution and FFU layout in the clean space. This significantly reduces the total system energy consumption while ensuring cleanliness standards are met, avoiding the energy waste associated with traditional equal-power operation or simple differential pressure control. This method is particularly suitable for cleanrooms, operating rooms, laboratories, and other scenarios with large areas, multiple FFUs, and variable loads.

[0091] In one embodiment, the spatial coordinated control unit 402 is further configured to:

[0092] Continuously monitor real-time cleanliness data in different areas of the space and generate a heat map of cleanliness distribution.

[0093] Based on the comparison between the cleanliness data at each location in the heat map and the target cleanliness, the space is divided into at least one cleanliness compliant area and at least one cleanliness non-compliant area.

[0094] Obtain the installation coordinates of each FFU in space;

[0095] For any area that meets the cleanliness standard and any adjacent area that does not meet the cleanliness standard, the FFU in the area that meets the cleanliness standard is the boundary FFU whose distance between the installation position and the boundary line between the two areas is less than a preset first threshold.

[0096] The boundary FFUs and all or part of the FFUs in the area where the cleanliness does not meet the standard are divided into the same coordinated control group. The rotation speed of the boundary FFUs in the coordinated control group is increased to guide the clean airflow to the area where the cleanliness does not meet the standard.

[0097] The spatial collaborative control unit 402 continuously collects real-time cleanliness data from each detection point at a fixed sampling period (e.g., once every 10 seconds) using multiple particulate matter concentration sensors (such as laser particle counters) deployed within the clean space. The unit correlates the position coordinates of each sensor with its readings and uses interpolation algorithms (such as inverse distance weighting or Kriging) to generate a continuous cleanliness distribution heatmap of the entire clean space. Different colors in the heatmap represent different cleanliness levels (e.g., green indicates better than the target, and red indicates worse than the target).

[0098] The real-time cleanliness level at each location (or each discrete grid point) on the heatmap is compared to a pre-defined target cleanliness level (e.g., ISO Class 5, i.e., 3,520 particles / m³):

[0099] Cleanliness standards met: Real-time value ≤ Target value;

[0100] Areas that do not meet cleanliness standards: Real-time value > Target value.

[0101] Using image segmentation or connected component analysis algorithms, the space is divided into several continuous compliant and non-compliant areas, and the boundary contours of each area are recorded. The geometric center or air outlet center of each FFU in the cleanroom plane coordinate system is read from the system configuration database (x, y). If the FFUs are arranged at different heights (e.g., ceiling, side wall), the vertical coordinates also need to be recorded.

[0102] For each pair of adjacent compliant and non-compliant areas (i.e., the two areas share a boundary line), perform the following operations:

[0103] Calculate the shortest distance from the installation location of each FFU within the compliance area to the boundary line;

[0104] If the distance is less than a preset first threshold (e.g., 0.5 meters, which can be adjusted according to the FFU air jet radius), then the FFU is marked as a boundary FFU. Note that a compliant area may be adjacent to multiple non-compliant areas simultaneously, and each must be handled separately.

[0105] For each pair of adjacent regions mentioned above, first determine the core members, which are the boundary FFUs determined in the previous step (located within the compliant region); then determine the extended members, which are all or some of the FFUs within the non-compliant region. All FFUs can be selected, or only some FFUs closer to the boundary line can be selected (e.g., the distance is less than the second threshold) to reduce unnecessary intervention.

[0106] The two FFUs are then combined into a coordinated control group. The system sends a command to the boundary FFUs within this coordinated control group to increase their rotation speed (or airflow) by a preset step (e.g., 5%, but not exceeding the upper limit). FFUs in non-compliant areas can remain unchanged, appropriately reduce their rotation speed, or wait for subsequent coordinated commands. After increasing the rotation speed of the boundary FFUs, the clean airflow jet velocity and pressure generated from the compliant area increase, which will "push" the adjacent non-compliant area horizontally or diagonally, thereby diluting or replacing the polluted air in that area.

[0107] The system continuously monitors the cleanliness distribution after the update. If, after an adjustment cycle (e.g., 30 seconds), the previously non-compliant area gradually becomes compliant, the boundary FFU speed is gradually reduced to return to energy-saving mode; if it still does not meet the standard, the boundary FFU speed can be iteratively increased, or the FFUs inside the non-compliant area can be further coordinated for adjustment.

[0108] Traditional cleanroom control often relies on return air pressure differential or overall air exchange rate, making it difficult to respond quickly to localized contamination hotspots. This embodiment utilizes boundary FFUs (Fan Filter Units) to actively supply air, forming a directional clean airflow barrier. This actively "squeezes" clean air from the compliant zone into the non-compliant zone, achieving rapid dilution and removal of contaminants. This embodiment adjusts only a small number of FFUs at the boundary of the compliant zone, rather than increasing the rotation speed of all FFUs, and avoids forcing all FFUs in the non-compliant zone to operate at full capacity. This precise coordination reduces unnecessary energy consumption, making it particularly suitable for scenarios with large areas and localized, instantaneous contamination (such as personnel movement or equipment door opening). Since the boundary FFUs are located within the compliant zone, increasing their rotation speed does not compromise the cleanliness of the compliant zone itself (because this area is already superior to the target); instead, it further strengthens its positive pressure state, preventing contaminants from the non-compliant zone from diffusing back into the compliant zone, thus providing dynamic isolation.

[0109] In one embodiment, the device self-learning compensation unit 403 is further configured to:

[0110] Record the standard speed-airflow characteristic curve of each FFU under factory calibration conditions;

[0111] Periodically obtain the current speed, current air volume, and current filter resistance value of the FFU, determine the actual speed-air volume characteristic curve, and calculate the deviation function between the actual speed-air volume characteristic curve and the standard speed-air volume characteristic curve.

[0112] By utilizing the trend of the deviation function over time, a speed compensation model is established and the speed compensation coefficient is solved.

[0113] The speed compensation coefficient is used to multiply the theoretical speed in the theoretical control command to obtain the corrected final control command.

[0114] First, under factory calibration conditions (i.e., the HEPA filter is brand new, the motor is not aged, and environmental conditions are standard), the airflow data of the FFU at different speed levels (or inverter frequencies) is measured using a wind tunnel or standard anemometer. This data is then fitted into a standard characteristic curve and stored in the non-volatile memory of the equipment's self-learning compensation unit. This curve serves as the benchmark for subsequent comparisons. The system automatically reads the following three parameters for each FFU at preset intervals (e.g., every 1 hour or every 24 hours):

[0115] Current speed: obtained through inverter feedback signal or Hall sensor;

[0116] Current air volume: The volumetric flow rate is obtained by integrating the air velocity sensor (array) installed in the FFU outlet or static pressure box;

[0117] Current filtration resistance: Measured in real time by differential pressure transmitters installed upstream and downstream of the HEPA filter.

[0118] Based on multiple recently accumulated data points (speed, airflow) (e.g., the last 10 samples), a least-squares fitting method is used to generate the actual characteristic curve for the current stage. Due to filter clogging and motor aging, this curve is usually located below the standard curve. At the same speed, the difference (or relative difference) between the actual airflow and the standard airflow is calculated to obtain the deviation function. This deviation function gradually increases with operating time and changes in filter resistance.

[0119] To compensate the actual air volume back to the standard air volume level, a speed compensation coefficient is introduced. This ensures that the final command speed = theoretical speed × This coefficient varies with running time. Dynamic adjustment. Preferably, establishing the speed compensation model and solving for the speed compensation coefficient includes:

[0120]

[0121] In the formula, This is the speed compensation coefficient. for Real-time filtering resistance at all times This is the initial filter resistance at the factory. This refers to the standard airflow at standard rotation speed. To run for a period of time The actual measured air volume changes; The weighting factor for filter clogging and aging is... This is the weighting coefficient for motor performance degradation.

[0122] In the above model, Dimensionless, with an initial value of 1 (i.e., no compensation), and becomes greater than 1 over time or as performance decays. This is the initial resistance at the factory, which is the resistance value of the new filter at the rated airflow. This is the cumulative effect of filter clogging. It represents the integral of the excess resistance relative to the initial value over time from the start of operation to the current moment. Since filter clogging is an irreversible cumulative process, this integral value will continue to increase with operating time. The principle is that for every percentage increase in resistance, a corresponding speed compensation is needed to maintain airflow, and this demand accumulates. The filter clogging and aging weighting coefficient is used to adjust the influence of the integral term on the compensation coefficient, and is usually calibrated experimentally (e.g., between 0.01 and 0.1). The air volume is usually taken at the rated speed. This refers to the actual airflow measured during current operation at the same standard operating speed. Due to motor performance degradation, Usually less than ; This is the relative deviation term for motor performance degradation. It reflects the proportion of efficiency reduction in the motor (or impeller, bearing) due to wear and aging. This deviation does not depend on integral accumulation but is a direct measurement of the current state. This is the motor performance degradation weighting coefficient, used to adjust the influence of the current deviation term on the compensation coefficient (e.g., 0.5 to 2.0).

[0123] The speed compensation coefficient consists of three parts: a base value of 1 (no compensation), plus the additional compensation requirement accumulated over time due to filter clogging, and plus the immediate compensation requirement caused by the current decrease in motor efficiency. The two parts correspond to different physical mechanisms and can be adjusted independently.

[0124] In calculation At that time, start from the first run Initially, the current resistance is recorded in each sampling period. Calculate relative excess Then, numerical integration (such as the trapezoidal rule) is used to accumulate the result to the current moment. This integral value monotonically does not decrease. A "standard speed verification test" is performed periodically (e.g., every 24 hours), by adjusting the FFU to the standard speed (e.g., 80% of the rated speed) and measuring the actual airflow at this time. and the standard air volume at that speed. The relative deviations are then obtained through comparison. Next, the integral term is multiplied by a weighting factor. Motor attenuation term multiplied by weighting factor Add 1 to get the current compensation coefficient. To avoid sudden changes, an exponential moving average is used. Filtering is performed to output a stable compensation coefficient.

[0125] When the upper-level control unit (such as a multi-parameter fusion unit or spatial collaborative control unit) calculates the theoretical speed of a certain FFU, the equipment self-learning compensation unit multiplies it by the current speed compensation coefficient, and then sends the final command to the FFU driver. In this way, even if filter clogging or motor aging causes a decrease in airflow, the system can automatically increase the actual speed to output the desired clean airflow. Simultaneously, the system has self-learning and update functions, allowing the compensation coefficient to dynamically adjust according to changes in equipment status. It can also record... The trend of change over time, when When the speed exceeds the preset limit (e.g., 1.35, which means an additional 35% speed increase is required), a warning will be issued to replace the filter or repair the motor.

[0126] The model described above separates the cumulative effect of filter clogging (related to time integral) from the instantaneous degradation of motor performance (related to current airflow deviation), allowing for separate calibration of weighting coefficients and more accurately reflecting different aging mechanisms. Simultaneously, the integral term inherently possesses a "memory" function, ensuring that even short-term resistance fluctuations (such as slight changes in resistance caused by humidity variations) will not cause drastic jumps in the compensation coefficient, thus guaranteeing control stability.

[0127] In one embodiment, the set of FFU performance parameters acquired by the parameter input module 20 includes the speed regulation response lag time constant of each FFU.

[0128] The AI ​​intelligent control module 40 also includes a feedforward prediction control unit 404, such as Figure 2 As shown, specifically, the feedforward prediction control unit 404 is used for:

[0129] When adjusting the speed or controlling the coordinated control group, based on the speed regulation response lag time constant, a pre-overshoot control command higher than the final control command value is output before outputting the final control command value.

[0130] After a preset time associated with the speed regulation response lag time constant, the control command is switched to the final control command value to shorten the response time for the cleanliness to reach the target value.

[0131] The parameter input module pre-acquires the speed regulation response lag time constant for each FFU (i.e., the delay time required from the issuance of the command to the actual speed / airflow reaching the target value). When the feedforward predictive control unit in the AI ​​intelligent control module needs to perform speed adjustment or coordinated control group regulation, it does not directly send the final target command, but first outputs a "pre-overshoot control command" higher than the target value. The amplitude of this pre-overshoot command is internally calculated to drive the motor to accelerate rapidly within the lag time. After experiencing a preset delay associated with this lag time constant (e.g., equal to or slightly less than the lag time constant), the control unit automatically switches the command back to the final target command value. At this time, since the motor has obtained sufficient acceleration kinetic energy in the pre-overshoot stage, the actual speed smoothly transitions to near the target value, effectively offsetting the response delay caused by mechanical inertia.

[0132] This solution enables FFUs to respond more quickly to changes in cleanliness requirements, significantly shortening the response time from the issuance of control commands to the achievement of target cleanliness levels. It also avoids the repeated overshoot and oscillations caused by lag in traditional PID control, improving cleanliness recovery speed while reducing unnecessary speed fluctuations and energy consumption. It is particularly suitable for scenarios with sudden changes in cleanliness (such as personnel entry or equipment door opening), effectively preventing the spread of contaminants.

[0133] In one embodiment, the data acquisition module 30 further includes at least one particle size classification particle counter, deployed at the air outlet of the FFU, for collecting particulate matter concentration data at at least two different particle size levels.

[0134] The AI ​​intelligent control module 40 also includes a target dynamic adaptation unit 405, such as Figure 2 As shown, the target dynamic adaptation unit 406 is used for:

[0135] The particulate matter concentration data at different particle size levels are compared one by one with the threshold values ​​of the corresponding particle size levels in the target cleanliness level.

[0136] Based on the comparison results, the dominant particle size factor that currently hinders the achievement of cleanliness standards was determined;

[0137] Based on the dominant particle size factor, the control instructions output by the AI ​​intelligent control module are adjusted to prioritize reducing the particle concentration corresponding to the dominant particle size factor.

[0138] The data acquisition module deploys at least one particle counter at the outlet of each FFU (Fan Filter Unit), capable of collecting particulate matter concentration data for two or more different particle size ranges (e.g., 0.3μm, 0.5μm, 1.0μm, etc.). The target dynamic adaptation unit in the AI ​​intelligent control module compares the real-time concentration of each particle size range with the corresponding particle size threshold in the target cleanliness level (e.g., ISO 14644 standard). By analyzing the degree to which each range exceeds or approaches the standard, it determines the main particulate matter size range hindering the achievement of cleanliness standards, i.e., the "dominant particle size factor" (e.g., if 0.3μm particles are severely excessive while 0.5μm is within the standard, then the dominant factor is small particle size). Based on this determination, the target dynamic adaptation unit dynamically adjusts the control commands output by the AI ​​intelligent control module, such as prioritizing increasing the rotation speed of FFUs with higher filtration efficiency or adjusting airflow organization to accelerate the discharge of small particle size particles, thereby specifically reducing the particulate matter concentration corresponding to the dominant particle size factor.

[0139] This solution enables intelligent identification and differentiated response to cleanroom contamination types. Different processes generate contaminants with varying particle size distributions (e.g., photolithography easily produces submicron particles, while grinding easily produces large particles), making traditional uniform control methods inefficient. This solution, by locking onto the dominant particle size factor and prioritizing control of key contamination sources, significantly improves the timeliness and energy efficiency of achieving cleanliness standards. Simultaneously, it avoids over-adjustment of already compliant particle size levels, reducing unnecessary energy consumption. It is particularly suitable for complex clean environments where multiple particle size contaminants coexist.

[0140] See Figure 3-4 The document provides an overall structural block diagram of the AI-based FFU group control cabinet linkage control system and an internal structural block diagram of the AI ​​intelligent control module. For example... Figure 3 As shown, in one embodiment, the system includes a group control cabinet (serving as the main carrier, internally integrating a parameter input module, a display module, an AI intelligent control module, a communication module, and a power supply module), a data acquisition module, an FFU linkage module, and FFUs; wherein, the AI ​​intelligent control module is bidirectionally connected to the parameter input module, the display module, the communication module, the power supply module, and the FFU linkage module; the communication module is bidirectionally connected to the data acquisition module and the FFU linkage module; the FFU linkage module is bidirectionally connected to the FFUs; the power supply module is unidirectionally connected to all modules (labeled "Power Supply"); the data acquisition module is labeled "Distributed in the clean space and FFUs".

[0141] The main body of the group control cabinet serves as the installation carrier and operation terminal of the system, integrating parameter input module, display module, AI intelligent control module, communication module and power supply module. It adopts a dustproof, flame-retardant and anti-interference shell design, which is suitable for the harsh working conditions of industrial clean rooms. It can be fixedly installed in a designated location in the workshop for easy operation and maintenance by users. The main body of the group control cabinet is equipped with wiring channels and heat dissipation structure to avoid overheating of internal components or signal interference, ensuring long-term stable operation of the system.

[0142] The parameter input module is located on the control panel of the main unit of the group control cabinet. It is used by users to input various basic parameters required for control, including but not limited to: FFU performance parameters (rated speed, rated air volume, power, filtration efficiency, and speed range of a single FFU), spatial parameters (length, width, and height of the clean space, i.e., space volume), FFU configuration parameters (number of FFUs, installation location distribution, and serial number of each FFU), and target cleanliness parameters (such as cleanliness levels of 100, 1000, and 10,000, or specific particulate matter concentration thresholds). Input methods support touch input and button input, and also support parameter import (such as importing preset parameters via USB flash drive). Input parameters can be stored in real time in the storage unit of the AI ​​intelligent control module for easy subsequent retrieval and modification, and have parameter encryption function to prevent accidental modification.

[0143] The AI ​​intelligent control module, as the core control unit of the system, is electrically connected to the parameter input module, data acquisition module, FFU linkage module, communication module, and display module. It has built-in AI intelligent algorithms, data storage unit, and logic decision-making unit. Its core functions include: parameter parsing, AI algorithm operation, logic decision-making, data storage and retrieval.

[0144] The data acquisition module consists of multiple sensors distributed in different areas of the cleanroom (including the interior of the space, FFU outlets, return air areas) and on each FFU. These sensors are used to collect various operational data in real time, including: real-time cleanliness data (particulate matter concentration, number of suspended particles, and particle detection for different particle sizes), FFU operational data (actual rotation speed, airflow, power, operating time, and filter resistance for each FFU), and environmental auxiliary data (temperature, humidity, and pressure difference within the cleanroom). The collected data is transmitted in real time to the AI ​​intelligent control module via a communication module, providing real-time data support for AI algorithm calculations and logical decisions. The sensors employ a high-precision, anti-interference design to ensure the accuracy of the collected data, and also have a self-calibration function to periodically calibrate the acquisition accuracy and avoid data deviation.

[0145] The FFU linkage module is electrically connected to each FFU and receives control commands from the AI ​​intelligent control module. This enables independent control of each FFU and coordinated control of multiple FFUs. Specific functions include: controlling the start and stop of FFUs, adjusting FFU speed (achieving stepless airflow adjustment), and providing real-time operating status feedback for each FFU to the AI ​​intelligent control module. It also supports group control of multiple FFUs, allowing FFUs to be divided into multiple groups based on their installation location and cleanroom area division. The AI ​​intelligent control module can send differentiated control commands to different groups to address uneven cleanliness in specific areas. Furthermore, it features a fault self-diagnosis function, promptly reporting any malfunction to the AI ​​intelligent control module for quick troubleshooting and maintenance.

[0146] The communication module employs a combination of wired and wireless communication to achieve data transmission and command interaction between modules, including communication between the main control cabinet and each FFU, and communication between the data acquisition module and the AI ​​intelligent control module. Wired communication uses RS485 bus or Ethernet to ensure the stability and speed of data transmission. Wireless communication uses LoRa technology to reduce wiring workload, improve system deployment efficiency, and has anti-interference capabilities to avoid electromagnetic interference affecting data transmission in industrial environments. The communication module supports automatic networking, which can construct a global topology map of FFU nodes by sending topology detection signals, realizing automated deployment and address allocation of FFUs and reducing networking difficulty.

[0147] The display module is located on the main control panel of the group control cabinet. It uses a high-definition touch screen to display the following information in real time: input basic parameters, real-time cleanliness data of the clean space, real-time operating parameters of each FFU (speed, air volume, power, etc.), calculation results of the AI ​​intelligent control module, and system operating status (normal, alarm). It supports users to query historical data, adjust parameters, and manually intervene in the FFU operating status (manual and automatic modes can be switched). When an abnormal situation occurs (such as non-compliance with cleanliness standards or FFU failure), the display module will issue text and light alarm prompts and simultaneously feed back to the AI ​​intelligent control module to trigger corresponding processing instructions.

[0148] The power supply module provides a stable power supply for the entire system, including the main modules of the group control cabinet, the data acquisition module, the FFU linkage module, and the power supply for each FFU. It has a built-in voltage regulator and backup power supply. When the external power supply is interrupted, the backup power supply can be automatically activated to ensure the short-term operation of the core modules of the system (AI intelligent control module, data acquisition module) and key FFUs, avoid a sudden drop in cleanliness, and prevent data loss. The power supply module has overload, short circuit, and overvoltage protection functions to ensure the system's power safety and adapt to industrial power supply environments.

[0149] The auxiliary function modules can be optionally equipped with a remote monitoring module and an alarm classification module: The remote monitoring module allows users to remotely view the system's operating status, adjust parameters, and receive alarm information via mobile phones, computers, and other terminals, enabling remote operation and maintenance; the alarm classification module classifies alarm information into three levels: alert, warning, and emergency shutdown. For example, FFU communication interruption belongs to the emergency shutdown level, and abnormal speed of a single FFU belongs to the alert level, which helps maintenance personnel quickly distinguish fault priorities and improve maintenance efficiency.

[0150] like Figure 4 As shown, the AI ​​intelligent control module also includes an AI deep learning model, a data storage unit, and a logic decision-making unit. The AI ​​deep learning model is bidirectionally connected to both the data storage unit and the logic decision-making unit, and the logic decision-making unit is bidirectionally connected to the data storage unit. The AI ​​deep learning model is labeled "receive parameters + real-time data, calculate optimal parameters", the data storage unit is labeled "store various parameters and operating data", and the logic decision-making unit is labeled "generate control / alarm commands".

[0151] See Figure 5-6 , Figure 5 A structural diagram of the main body of a group control cabinet is provided. The main body of the group control cabinet is a cuboid, with an operation panel on the front. Above the panel is a high-definition touch screen (labeled "display module"), and below the screen is a touch / button input area (labeled "parameter input module"). Below the input area are a power switch and indicator lights (labeled "run indicator light" and "alarm indicator light"). The sides of the group control cabinet are labeled "wiring channel" and "heat dissipation vent", and the bottom is labeled "fixed mounting holes". The overall dimensions of the group control cabinet are marked (standard dimensions can be marked, such as 80cm×60cm×120cm). Figure 6 It provides a grouping diagram for FFU linkage control. The clean space plan is marked with the length and width dimensions (standard dimensions can be marked, such as 10m×8m), dividing the space into three areas: front, middle, and back (or as needed). Several circles are drawn in each area (representing FFUs), and the group number (such as group 1, group 2, group 3) and the number of FFUs in each group are marked. Each group is marked with an arrow pointing to the main body of the group control cabinet and labeled "linkage control". Several sensor locations are marked in the space (circles are marked "sensor") and labeled "data acquisition".

[0152] The front of the control cabinet features a high-definition touchscreen display and a touch / button input area, providing a clear view of system operating status, parameter configurations, and alarm information. It also allows on-site personnel to quickly input or modify control parameters, lowering the operational threshold. The cabinet's sides are equipped with wiring channels and ventilation vents, while the bottom features mounting holes, ensuring neat and orderly internal cabling, unobstructed heat dissipation, and secure mounting with anchor bolts, adapting to the installation requirements of cleanrooms and other industrial environments. The cleanroom space is divided into front, middle, and rear zones based on its dimensions, with each FFU group independently numbered and linked to the control cabinet for coordinated control. The system can perform differentiated control based on real-time cleanliness differences in each zone, avoiding energy waste caused by uniform global adjustments and shortening the response time to localized contamination.

[0153] Multiple sensor nodes are deployed within the space to collect parameters such as cleanliness and wind speed in each area in real time and upload them to the group control cabinet. Combined with a group control strategy, it can accurately identify pollution hotspots and dynamically optimize the operating status of each FFU group, achieving on-demand air supply. The front of the group control cabinet is equipped with operation and alarm indicator lights, allowing maintenance personnel to quickly determine whether the system is functioning normally or experiencing a fault. Combined with detailed logs on the touchscreen, this facilitates rapid problem location and maintenance.

[0154] See Figure 7-8 In one embodiment of the present invention, a method for FFU group control and linkage regulation based on multi-parameter fusion and self-learning mechanism is also provided, applied to the system described in any of the foregoing embodiments, comprising the following steps:

[0155] S10. Obtain static configuration parameters input by the user and collect dynamic running parameters in real time;

[0156] S20. Associate and map static configuration parameters with dynamic operating parameters to construct an initial digital twin model of the clean space and calculate the theoretical basic control commands for each FFU.

[0157] S30. Based on the spatial cleanliness distribution gradient, all FFUs are dynamically divided into multiple collaborative control groups to generate differentiated group control instructions.

[0158] S40. Based on the running time, filter resistance change and response deviation of each FFU, call its personalized aging compensation model, modify the basic control command and group control command for each unit, and generate the final control command.

[0159] S50. Based on the final control command, perform independent drive control on each FFU or each coordinated control group.

[0160] First, static configuration parameters input by the user (such as spatial geometry, FFU layout, and target cleanliness level) are acquired, and dynamic operating parameters (such as particle concentration, air velocity, and filtration resistance) are collected in real time. Then, these two types of parameters are correlated and mapped to construct a digital twin model of the clean space, calculating the theoretical basic control commands for each FFU. Next, based on the spatial cleanliness distribution gradient, all FFUs are dynamically divided into multiple collaborative control groups, and differentiated group control commands are generated for different areas. Then, based on the runtime, filtration resistance changes, and response deviations of each FFU, a personalized aging compensation model is invoked to correct the basic commands and group commands for each FFU individually, eliminating the impact of equipment characteristic drift. Finally, the corrected final control commands are issued to each FFU or collaborative control group to achieve independent drive control.

[0161] Specifically, the system workflow sequence can be found in [reference needed]. Figure 8 The entire process is automated and requires no manual intervention (intervention modes can be manually switched), including the following steps:

[0162] 1. Parameter input: Users input basic parameters such as FFU performance parameters, space volume, number of FFUs, and target cleanliness through the parameter input module of the group control cabinet. The system automatically verifies the integrity of the parameters and prompts for missing or incorrect parameters. After user confirmation, the parameters are stored in the AI ​​intelligent control module.

[0163] 2. Initial startup: The user starts the system, the power supply module supplies power to each module, and the data acquisition module begins to collect the initial cleanliness data of the clean space and the initial operating data of the FFU in real time, which is then transmitted to the AI ​​intelligent control module through the communication module.

[0164] 3. AI Intelligent Calculation: The AI ​​intelligent control module combines the input basic parameters with real-time collected data, and uses an AI deep learning model to calculate the optimal operating parameters (speed, air volume) for each FFU, generating linkage control commands;

[0165] 4. Coordinated Control: The FFU linkage module receives control commands and controls each FFU to start and run according to the optimal operating parameters, realizing the coordinated linkage of multiple FFUs; at the same time, the data acquisition module continuously collects real-time data and feeds it back to the AI ​​intelligent control module in real time.

[0166] 5. Dynamic optimization: The AI ​​intelligent control module compares the collected cleanliness data with the target cleanliness parameters in real time. If there is a deviation (such as cleanliness exceeding or failing to meet the standard), or if the FFU operating data is abnormal (such as increased filtration resistance or speed deviation), the control command is automatically adjusted to optimize the operating parameters of each FFU to ensure that the cleanliness is stably up to standard, while achieving energy saving and consumption reduction.

[0167] 6. Status Feedback and Alarm: The display module shows the system's operating status and various data in real time. If the cleanliness level fails to meet the standard, the FFU fails, or the power supply is abnormal, the system will automatically issue an alarm and record the fault data for subsequent troubleshooting.

[0168] 7. Self-learning optimization: During long-term system operation, the AI ​​intelligent control module stores historical operating data and continuously optimizes the calculation logic through the AI ​​deep learning model to improve the control accuracy and energy-saving effect, and adapt to changes in different working conditions (such as FFU aging and cleanliness requirements adjustment).

[0169] This process balances global optimization with localized improvements and adapts to equipment aging, significantly reducing energy consumption while maintaining cleanliness. It is understood that the specific implementation and effects of the method provided in this embodiment can be found in the description of the system embodiment above, and for the sake of brevity, will not be repeated here.

[0170] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

Claims

1. A group control and linkage regulation system for FFUs based on multi-parameter fusion and self-learning mechanism, characterized in that, The system includes: The main body of the group control cabinet; The parameter input module is used to obtain static configuration parameters input by the user. The static configuration parameters include at least FFU performance parameters, clean space geometric parameters, FFU layout and quantity parameters, and target cleanliness level. The data acquisition module is used to acquire dynamic operating parameters in real time. The dynamic operating parameters include at least the cleanliness data of multiple points in the space, the real-time operating parameters and status data of each FFU. The AI ​​intelligent control module includes: The multi-parameter fusion unit is used to map static configuration parameters to dynamic operating parameters, build an initial digital twin model of the clean space, and calculate the theoretical control commands for each FFU. The spatial coordinated control unit is used to identify the cleanliness distribution gradient based on static configuration parameters and dynamic operating parameters, dynamically divide all FFUs into multiple coordinated control groups based on the identification results, and generate differentiated control instructions for different coordinated control groups. The equipment self-learning compensation unit is used to build a personalized aging compensation model for each FFU based on the running time, filter resistance change curve and speed-airflow response deviation of each FFU throughout its entire life cycle. Based on the aging compensation model, the theoretical control commands are corrected to generate individual compensation commands for each FFU. The FFU linkage module is used to receive differentiated control commands and individual compensation commands, determine the coordinated control group to which each FFU belongs based on the differentiated control commands, correct the benchmark control value of each FFU to which the coordinated control group belongs on a unit-by-unit basis based on the individual compensation commands, and output the final control command to independently drive and control each FFU or each coordinated control group.

2. The FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism according to claim 1, characterized in that, The multi-parameter fusion unit is also used for: Based on the geometric parameters of the clean space and the layout and number of FFUs, a spatial airflow organization influence matrix is ​​established. The matrix is ​​used to characterize the contribution weight of each FFU to the cleanliness of different locations in the space. The contribution weight is determined according to the physical distance and the obstacle obstruction factor. Based on the difference between the target cleanliness level and the cleanliness data of multiple points in the current space, and combined with the contribution weight, the theoretical rotation speed and air volume of each FFU are solved in reverse iteration to minimize the difference and optimize the total energy consumption of all FFUs.

3. The FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism according to claim 1, characterized in that, The space-coordinated control unit is also used for: Continuously monitor real-time cleanliness data in different areas of the space and generate a heat map of cleanliness distribution. Based on the comparison between the cleanliness data at each location in the heat map and the target cleanliness, the space is divided into at least one cleanliness compliant area and at least one cleanliness non-compliant area. Obtain the installation coordinates of each FFU in space; For any area that meets the cleanliness standard and any adjacent area that does not meet the cleanliness standard, the FFU in the area that meets the cleanliness standard is the boundary FFU whose distance between the installation position and the boundary line between the two areas is less than a preset first threshold. The boundary FFUs and all or part of the FFUs in the area where the cleanliness does not meet the standard are divided into the same coordinated control group. The rotation speed of the boundary FFUs in the coordinated control group is increased to guide the clean airflow to the area where the cleanliness does not meet the standard.

4. The FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism according to claim 1, characterized in that, The device self-learning compensation unit is also used for: Record the standard speed-airflow characteristic curve of each FFU under factory calibration conditions; Periodically obtain the current speed, current air volume, and current filter resistance value of the FFU, determine the actual speed-air volume characteristic curve, and calculate the deviation function between the actual speed-air volume characteristic curve and the standard speed-air volume characteristic curve. By utilizing the trend of the deviation function over time, a speed compensation model is established and the speed compensation coefficient is solved. The speed compensation coefficient is used to multiply the theoretical speed in the theoretical control command to obtain the corrected final control command.

5. The FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism according to claim 4, characterized in that, The process of establishing the speed compensation model and solving for the speed compensation coefficients includes: ； In the formula, This is the speed compensation coefficient. for Real-time filtering resistance at all times This is the initial filter resistance at the factory. This refers to the standard airflow at standard rotation speed. To run for a period of time The actual measured air volume changes; The weighting factor for filter clogging and aging is... This is the weighting coefficient for motor performance degradation.

6. The FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism according to claim 1, characterized in that, The FFU performance parameter set obtained by the parameter input module includes the speed regulation response lag time constant of each FFU. The AI ​​intelligent control module further includes a feedforward prediction control unit, which is used for: When adjusting the speed or controlling the coordinated control group, based on the speed control response lag time constant, a pre-overshoot control command higher than the final control command value is output before outputting the final control command value. After a preset time associated with the speed regulation response lag time constant, the control command is switched to the final control command value to shorten the response time for the cleanliness to reach the target value.

7. The FFU group control and linkage regulation system based on multi-parameter fusion and self-learning mechanism according to claim 1, characterized in that, The data acquisition module also includes at least one particle size classification particle counter, which is deployed at the air outlet of the FFU to collect particulate matter concentration data at at least two different particle size ranges. The AI ​​intelligent control module also includes a target dynamic adaptation unit, used for: The particulate matter concentration data at different particle size levels are compared one by one with the threshold values ​​of the corresponding particle size levels in the target cleanliness level. Based on the comparison results, the dominant particle size factor that currently hinders the achievement of cleanliness standards was determined; Based on the dominant particle size factor, the control instructions output by the AI ​​intelligent control module are adjusted to prioritize reducing the particle concentration corresponding to the dominant particle size factor.

8. A method for coordinated control of FFU groups based on multi-parameter fusion and self-learning mechanisms, applied to the system as described in any one of claims 1 to 7, characterized in that, Includes the following steps: Obtain static configuration parameters input by the user and collect dynamic runtime parameters in real time; By associating and mapping static configuration parameters with dynamic operating parameters, an initial digital twin model of the clean space is constructed, and the theoretical basic control commands for each FFU are calculated. Based on the spatial cleanliness distribution gradient, all FFUs are dynamically divided into multiple collaborative control groups, generating differentiated group control instructions. Based on the runtime, filter resistance changes, and response deviation of each FFU, its personalized aging compensation model is invoked to modify the basic control commands and group control commands for each unit, generating the final control commands. Based on the final control instructions, each FFU or each coordinated control group is driven and controlled independently.

Images