A method and system for optimizing intelligent industrial production environment based on historical data

Abstract

This invention relates to the field of environmental optimization technology, and in particular to an intelligent industrial production environment optimization method and system based on historical data, comprising the following steps: Step 1, constructing a historical database: recording historical data of the industrial production environment under different conditions, and establishing a benchmark model based on the historical database; Step 2, real-time data monitoring: real-time monitoring of the current suspended particulate concentration, current air exchange rate, current resistance and integrity status of the HEPA filter, and current airtightness status of the maintenance structure in the industrial production environment; Step 3, anomaly judgment; Step 4, activation of the suspended particulate concentration adjustment system; Step 5, model self-updating; This method incorporates the historical database through incremental learning or sliding window refitting methods to update the benchmark model, which is beneficial for automatically judging the output fault type. Through the feedback update mechanism, the system has learning capabilities and maintains optimal control performance over long-term operation.

Description

Technical Field

[0001] This invention relates to the field of environmental optimization, and in particular to an intelligent industrial production environment optimization method and system based on historical data. Background Technology

[0002] Industrial production environment optimization methods refer to a series of management activities that use systematic strategies and technical means to continuously improve various elements of the production site, aiming to improve efficiency, ensure safety, reduce waste, and ensure stable quality.

[0003] Current methods for controlling suspended particulate concentration in industrial production environments typically employ fixed concentration thresholds to trigger adjustment systems or rely solely on a single parameter for equipment start-up and shutdown. However, these methods struggle to distinguish the root causes of concentration anomalies, and fixed thresholds cannot adapt to dynamic drift caused by equipment aging or changes in operating conditions. This results in control lag, frequent malfunctions, and a lack of self-optimization capabilities. Therefore, there is an urgent need for an intelligent industrial production environment optimization method based on historical data. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for optimizing intelligent industrial production environments based on historical data.

[0005] This invention provides a method and system for optimizing intelligent industrial production environments based on historical data, comprising the following steps:

[0006] Step 1: Construct a historical database: Record historical data of industrial production environment under different conditions to form a historical database, and establish a benchmark model based on the historical database;

[0007] Step 2: Real-time data monitoring: Real-time monitoring of the current suspended particulate concentration, current air exchange rate, current resistance and integrity status of the HEPA filter, and current airtightness status of the maintenance structure in the industrial production environment to generate real-time monitoring data;

[0008] Step 3, Anomaly Judgment: Compare the real-time monitoring data with the benchmark model to calculate the suspended particle concentration deviation rate, filter resistance deviation rate, and airtightness deviation rate respectively. Based on the calculation results of each deviation rate, trace the source of the anomaly, determine the dominant cause of the current suspended particle concentration anomaly, and generate the reasoning result.

[0009] Step 4: Start-up of the suspended particle concentration regulation system: Based on the reasoning results, dynamically determine the start-up threshold of the suspended particle concentration regulation system, and select the corresponding control strategy to control the start-up of the suspended particle concentration regulation system.

[0010] Step 5: Model self-update: After the suspended particulate concentration adjustment system is started, monitor the concentration change trend to evaluate the control effect, and feed the data of this adjustment process back to the historical database to update the baseline model;

[0011] The system first collects a large amount of historical data during normal operation, covering suspended particle concentrations at different air change rates, the variation of filter resistance over operating time, and the attenuation trend of building envelope airtightness over usage time. It then establishes concentration-air change rate relationship models, filter resistance attenuation models, and airtightness attenuation models using statistical modeling or machine learning methods, providing a quantitative reference system for subsequent anomaly detection. The system deploys multiple types of sensors, including a laser particle counter to monitor concentration, a wind speed sensor to calculate air change rates, a differential pressure sensor to monitor the pressure difference between filter resistance and building envelope, and an online aerosol detection device to monitor filter integrity. These sensors continuously collect data using high-frequency sampling, forming a real-time data stream reflecting the current system status. The system substitutes the current air change rate into the concentration benchmark model to obtain the historical normal concentration that should exist at that air change rate, thereby calculating the concentration deviation rate and substituting the current filter operating time into the model. The system inputs the resistance attenuation model to obtain the expected resistance and calculates the resistance deviation rate. Substituting the current building envelope usage time into the airtightness attenuation model, it obtains the expected airtightness index and calculates the airtightness deviation rate. Through preset combination logic, the system attributes the concentration anomaly to one of the three main causes. The system dynamically determines the basic start-up threshold based on the concentration deviation rate distribution under historical normal operating conditions, making the start-up conditions adaptive with the operating conditions. At the same time, the threshold is compensated and adjusted according to the anomaly tracing results. If the filter or airtightness is abnormal, the start-up threshold is lowered to achieve early warning. After the control process ends, the system uses the parameter data of this event as a new sample and incorporates it into the historical database through incremental learning or sliding window refitting methods to update the baseline model. Through the combination judgment of multiple parameters, it is beneficial to automatically determine the output fault type, thereby shortening the fault judgment time. Through the feedback update mechanism, the system has the ability to learn and maintain optimal control performance in long-term operation.

[0012] Preferably, step one specifically includes:

[0013] The historical database consists of historical data on suspended particle concentration in industrial production environments at different air exchange rates, historical data on the resistance and integrity tests of high-efficiency filters, and historical data on the airtightness tests of maintenance structures.

[0014] The benchmark models include a benchmark relationship model between air exchange rate and suspended particle concentration, a benchmark model for filter resistance attenuation, and a benchmark model for airtightness attenuation.

[0015] Preferably, step three specifically includes:

[0016] The method for calculating the suspended particle concentration deviation rate is as follows:

[0017]

[0018] in, This represents the current real-time concentration of suspended particles. The average historical normal concentration value corresponding to the current number of air exchanges is obtained from the baseline model.

[0019] Preferably, step three specifically includes:

[0020] The filter resistance deviation rate is calculated as follows:

[0021]

[0022] in, The resistance of the high-efficiency filter is currently being monitored in real time. This is the expected resistance predicted based on the filter resistance attenuation benchmark model, corresponding to the current cumulative operating time of the filter.

[0023] Preferably, step three specifically includes:

[0024] The airtightness deviation rate is calculated as follows:

[0025]

[0026] in, The airtightness index of the maintenance structure currently under real-time monitoring is defined as either the pressure differential decay rate or the leakage rate per unit area. It refers to the historical benchmark airtightness index obtained from the airtightness attenuation benchmark model, which corresponds to the current maintenance structure's usage time or state.

[0027] Preferably, the specific method for anomaly tracing in step three is as follows:

[0028] When the suspended particle concentration deviation rate exceeds a set first threshold and the filter resistance deviation rate exceeds a set second threshold, the primary cause is determined to be abnormal integrity of the high-efficiency filter.

[0029] When the suspended particle concentration deviation rate exceeds the set first threshold and the airtightness deviation rate exceeds the set third threshold, the main cause is determined to be a decrease in the airtightness of the maintenance structure.

[0030] When the suspended particle concentration deviation rate exceeds the set first threshold, while the filter resistance deviation rate and air tightness deviation rate do not exceed the corresponding thresholds, the primary cause is determined to be insufficient air exchange rate.

[0031] Preferably, the control strategy in step four specifically includes:

[0032] When the result of the anomaly tracing is that the current air exchange rate is insufficient, the variable frequency fan is controlled to increase the operating frequency and increase the air exchange rate;

[0033] When the result of the anomaly tracing is that the integrity of the high-efficiency filter is abnormal, the control will issue an alarm prompt to replace or repair the high-efficiency filter, and temporarily put the backup filter unit into operation or increase the air supply volume in the adjacent area.

[0034] When the result of the anomaly tracing is that the airtightness of the building envelope has decreased, the control will issue an alarm for building envelope maintenance and temporarily increase the air supply to compensate for the pollution risk caused by the leak.

[0035] Preferably, step four specifically includes:

[0036] A1. Obtain the suspended particle concentration deviation rate by comparing the real-time monitoring data with the benchmark model;

[0037] A2. Extract the historical concentration deviation rate samples of suspended particles under normal operating conditions from the historical database, calculate the set percentile of the historical deviation rate distribution of suspended particles, and use it as the start threshold.

[0038] A3. Based on the anomaly tracing results, dynamically adjust the activation threshold:

[0039] If the anomaly tracing result indicates an abnormality in the integrity of the high-efficiency filter, then the activation threshold is reduced to the first adjustment value;

[0040] If the result of the abnormality tracing is a decrease in the airtightness of the maintenance structure, then the activation threshold is reduced to the second adjustment value;

[0041] If the abnormality tracing result indicates insufficient air exchange rate, then the activation threshold remains unchanged;

[0042] A4. Obtain the real-time change rate of suspended particle concentration. When the suspended particle concentration deviation rate exceeds the adjusted start threshold, and the concentration change rate exceeds the preset trend threshold or the duration of exceeding the standard exceeds the preset duration, the suspended particle concentration adjustment system is triggered to start.

[0043] Preferably, step five specifically includes:

[0044] Incremental learning or sliding window refitting methods are used to add the initial parameters, control actions, adjustment process and final effect of this adjustment process as new samples to the historical database, and update each model periodically or triggeredly.

[0045] A smart industrial production environment optimization system based on historical data includes:

[0046] The sensing module is used to collect data on suspended particle concentration, air exchange rate, filter resistance, filter integrity status, and airtightness of the maintenance structure.

[0047] The data storage module is used to store historical operating data and benchmark models;

[0048] The inference comparison module is used to compare real-time data with a benchmark model and perform anomaly tracing.

[0049] The control decision module is used to dynamically determine the start-up threshold and generate control commands;

[0050] The execution module is used to receive control commands and adjust the fan, valves and local purification unit;

[0051] The self-optimization module is used to evaluate the control effect and update the baseline model.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] By incorporating historical databases through incremental learning or sliding window refitting methods, the baseline model is updated. Through the combination of multiple parameters, it is beneficial to automatically determine the type of fault output, thereby shortening the fault determination time. Through the feedback update mechanism, the system has the ability to learn and maintain optimal control performance in long-term operation. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0055] Figure 2 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0056] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0057] like Figure 1 The method and system for optimizing an intelligent industrial production environment based on historical data, as shown, include the following steps:

[0058] Step 1: Construct a historical database: Record historical data of industrial production environment under different conditions to form a historical database, and establish a benchmark model based on the historical database;

[0059] Step 2: Real-time data monitoring: Real-time monitoring of the current suspended particulate concentration, current air exchange rate, current resistance and integrity status of the HEPA filter, and current airtightness status of the maintenance structure in the industrial production environment to generate real-time monitoring data;

[0060] Step 3, Anomaly Judgment: Compare the real-time monitoring data with the benchmark model to calculate the suspended particle concentration deviation rate, filter resistance deviation rate, and airtightness deviation rate respectively. Based on the calculation results of each deviation rate, trace the source of the anomaly, determine the dominant cause of the current suspended particle concentration anomaly, and generate the reasoning result.

[0061] Step 4: Start-up of the suspended particle concentration regulation system: Based on the reasoning results, dynamically determine the start-up threshold of the suspended particle concentration regulation system, and select the corresponding control strategy to control the start-up of the suspended particle concentration regulation system.

[0062] Step 5: Model self-update: After the suspended particulate concentration adjustment system is started, monitor the concentration change trend to evaluate the control effect, and feed the data of this adjustment process back to the historical database to update the baseline model;

[0063] The current method for controlling suspended particulate concentration in industrial production environments typically uses a fixed concentration threshold to trigger the adjustment system or relies on a single parameter to start and stop the equipment. However, such methods are difficult to distinguish the root cause of concentration anomalies, and fixed thresholds cannot adapt to dynamic drift caused by equipment aging and changes in operating conditions, resulting in control lag, frequent malfunctions, and a lack of self-optimization capabilities. Therefore, there is an urgent need for an intelligent industrial production environment optimization method based on historical data.

[0064] This embodiment of the invention can solve the above problems. The specific implementation is as follows: First, during normal operation, the system collects a large amount of historical data, covering suspended particle concentrations at different air change rates, the variation of filter resistance with operating time, and the attenuation trend of building envelope airtightness with usage time. Furthermore, it establishes concentration-air change rate relationship models, filter resistance attenuation models, and airtightness attenuation models through statistical modeling or machine learning methods, providing a quantitative reference system for subsequent anomaly detection. The system deploys multiple types of sensors, including a laser particle counter to monitor concentration, a wind speed sensor to calculate air change rates, a differential pressure sensor to monitor filter resistance and building envelope pressure difference, and an online aerosol detection device to monitor filter integrity. These sensors continuously collect data using high-frequency sampling, forming a real-time data stream reflecting the current system state. The system substitutes the current air change rate into the concentration benchmark model to obtain the historical normal concentration that should exist at that air change rate, thereby calculating the concentration deviation. The system calculates the resistance deviation rate by substituting the current filter operating time into the resistance attenuation model, obtaining the expected resistance, and calculating the resistance deviation rate by substituting the current building envelope usage time into the airtightness attenuation model, obtaining the expected airtightness index, and calculating the airtightness deviation rate. Through preset combination logic, the concentration anomaly is attributed to one of the three main causes. The system dynamically determines the basic start-up threshold based on the concentration deviation rate distribution under historical normal operating conditions, making the start-up conditions adaptive with the operating conditions. At the same time, the threshold is compensated and adjusted according to the anomaly tracing results. If the filter or airtightness is abnormal, the start-up threshold is lowered to achieve early warning. After the control process ends, the system uses the parameter data of this event as a new sample and incorporates it into the historical database through incremental learning or sliding window refitting methods to update the baseline model. Through the combination judgment of multiple parameters, it is beneficial to automatically determine the output fault type, thereby shortening the fault judgment time. Through the feedback update mechanism, the system has the ability to learn and maintain optimal control performance in long-term operation.

[0065] As an optional embodiment, step one specifically includes:

[0066] The historical database consists of historical data on suspended particle concentration in the industrial production environment at different air exchange rates, historical data on the resistance and integrity tests of high-efficiency filters, and historical data on the airtightness tests of the maintenance structure.

[0067] The benchmark models include a benchmark relationship model between air exchange rate and suspended particle concentration, a benchmark model for filter resistance attenuation, and a benchmark model for air tightness attenuation.

[0068] Historical data on suspended particle concentration at the same air change rate can record the distribution of suspended particle concentration under different air change rates, such as 10 times / h, 15 times / h, 20 times / h, etc., forming a mapping relationship between concentration and air change rate. This can characterize the dilution capacity of the clean environment and serve as a basic reference for judging whether the concentration is abnormal. Historical data on the resistance of HEPA filters and historical data on integrity tests record the resistance growth curve of the filter from new installation to disposal, as well as the results of periodic integrity tests, forming a resistance-time decay model. This can characterize the resistance change law during the normal aging process of the filter and serve as a basic reference for judging whether the filter has abnormal leakage or premature blockage. Historical data on the airtightness test of the building envelope records the pressure differential decay rate or leakage rate of the building envelope at different usage time points, forming an airtightness-time decay model. This can characterize the airtightness change law during the normal aging process of the building envelope and serve as a basic reference for judging whether the airtightness has abnormally decreased. This provides a clear data source for subsequent deviation rate calculations and avoids model failure caused by data chaos.

[0069] As an optional embodiment, step three specifically includes:

[0070] The method for calculating the suspended particle concentration deviation rate is as follows:

[0071]

[0072] in, This represents the current real-time concentration of suspended particles. The average historical normal concentration corresponding to the current number of air changes is obtained from the baseline model;

[0073] By introducing a comparison with the same number of air changes, the interference of changes in the number of air changes is eliminated, and the judgment of abnormal concentration is decoupled from the change in the number of air changes. This allows the deviation rate to truly reflect whether the concentration is abnormal under the same ventilation conditions, thus avoiding false alarms caused by production load adjustments.

[0074] As an optional embodiment, step three specifically includes:

[0075] The filter resistance deviation rate is calculated as follows:

[0076]

[0077] in, The resistance of the high-efficiency filter is currently being monitored in real time. The expected resistance, predicted based on the filter resistance attenuation benchmark model, is the resistance corresponding to the current cumulative operating time of the filter.

[0078] During normal use, the resistance of a high-efficiency filter gradually increases as the amount of dust collected rises. Therefore, the resistance value alone cannot directly indicate an anomaly. By comparing the current resistance with the expected resistance predicted based on usage time, if the measured resistance is significantly lower than expected, it suggests potential damage or leakage; if it is significantly higher than expected, it suggests possible premature clogging or improper selection. Through time-dimensional modeling, both excessively high and excessively low resistance anomalies can be identified simultaneously. When the resistance begins to deviate from the expected curve, even before reaching the final resistance threshold, the system can issue an early warning, preventing sudden filter failure and environmental out of control. This facilitates accurate assessment of the degree of clogging, avoiding premature or delayed replacement, and optimizing the filter replacement cycle.

[0079] As an optional embodiment, step three specifically includes:

[0080] The airtightness deviation rate is calculated as follows:

[0081]

[0082] in, The airtightness index of the maintenance structure currently under real-time monitoring is defined as either the pressure differential decay rate or the leakage rate per unit area. These are historical benchmark airtightness indices obtained from the airtightness attenuation benchmark model, corresponding to the current maintenance structure's usage time or state.

[0083] The airtightness of the building envelope will naturally decrease over time. When the rate of decrease in airtightness significantly exceeds the historical baseline, it indicates an abnormal leak. Using the expected airtightness as a benchmark helps the system distinguish between normal aging and abnormal failures. Dynamic benchmarks help avoid missed and false alarms. Furthermore, the airtightness deviation rate directly quantifies the severity of the leak, providing data support for maintenance prioritization, enabling predictive maintenance, and avoiding cleanliness accidents caused by sudden airtightness failure.

[0084] As an optional embodiment, the specific method of anomaly tracing in step three is as follows:

[0085] When the suspended particle concentration deviation rate exceeds a set first threshold and the filter resistance deviation rate exceeds a set second threshold, the primary cause is determined to be abnormal integrity of the high-efficiency filter.

[0086] When the suspended particle concentration deviation rate exceeds the set first threshold and the airtightness deviation rate exceeds the set third threshold, the main cause is determined to be a decrease in the airtightness of the maintenance structure.

[0087] When the suspended particle concentration deviation rate exceeds the set first threshold, while the filter resistance deviation rate and air tightness deviation rate do not exceed the corresponding thresholds, the main cause is determined to be insufficient current air exchange rate.

[0088] If the filter fails, dusty air from outside will directly penetrate the filter and enter the clean area, causing the suspended particle concentration deviation rate to exceed the set first threshold. At the same time, the filter resistance will deviate from the expected value, causing the filter resistance deviation rate to exceed the set second threshold.

[0089] If the building envelope leaks, outside air will intrude through the gaps, causing the suspended particle concentration deviation rate to exceed the set first threshold. At the same time, the pressure differential attenuation rate will increase, causing the airtightness deviation rate to exceed the set third threshold.

[0090] If the number of air changes is insufficient, the dilution capacity will decrease, causing the suspended particle concentration deviation rate to exceed the set first threshold. However, if the filter and enclosure structure are normal, the filter resistance deviation rate and air tightness deviation rate will not exceed the corresponding threshold.

[0091] By designing the above three rules, the system can identify the source of pollution, allowing maintenance personnel to directly know the type of fault after receiving an alarm, thereby greatly improving the repair response speed. Furthermore, when multiple anomalies occur simultaneously, the system will identify the dominant cause, avoiding decision-making confusion.

[0092] It should be noted that the thresholds need to be optimized independently. The first, second, and third thresholds can be set independently based on the statistical distribution of historical data, without interfering with each other, so as to facilitate system tuning.

[0093] As an optional embodiment, the control strategy in step four specifically includes:

[0094] When the result of the anomaly tracing is that the current air exchange rate is insufficient, the variable frequency fan is controlled to increase the operating frequency and increase the air exchange rate;

[0095] When the result of the anomaly tracing is that the integrity of the high-efficiency filter is abnormal, the control will issue an alarm prompt to replace or repair the high-efficiency filter, and temporarily put the backup filter unit into operation or increase the air supply volume in the adjacent area.

[0096] When the result of the anomaly tracing is that the airtightness of the building envelope has decreased, the control will issue an alarm for building envelope maintenance and temporarily increase the air supply to compensate for the pollution risk caused by the leak.

[0097] Insufficient ventilation: Control the variable frequency fan to increase the operating frequency and increase the number of air exchanges, thereby reducing the concentration by enhancing the dilution capacity;

[0098] Filter malfunction: An alarm is issued prompting the replacement or repair of the high-efficiency filter, and a backup filter unit is temporarily put into operation or the air supply volume in the adjacent area is increased to avoid production interruption, demonstrating the high availability design of the system;

[0099] Decreased air tightness: Issues an alarm for building envelope maintenance and temporarily increases air supply to compensate for the pollution risk caused by leakage. Pressure compensation is achieved by increasing air supply to maintain positive pressure between the clean area and adjacent areas, thus inhibiting the intrusion of external pollution.

[0100] By adopting different strategies for different causes, the efficiency and effectiveness of problem-solving were improved.

[0101] As an optional embodiment, step four specifically includes:

[0102] A1. Obtain the suspended particle concentration deviation rate by comparing the real-time monitoring data with the benchmark model;

[0103] A2. Extract the historical concentration deviation rate samples of suspended particles under normal operating conditions from the historical database, calculate the set percentile of the historical deviation rate distribution of suspended particles, and use it as the start threshold.

[0104] A3. Based on the anomaly tracing results, dynamically adjust the activation threshold:

[0105] If the result of the anomaly tracing is that the high-efficiency filter is not intact, then the start threshold is reduced to the first adjustment value;

[0106] If the result of the anomaly tracing is a decrease in the airtightness of the maintenance structure, then the activation threshold is reduced to the second adjustment value;

[0107] If the abnormality tracing result indicates insufficient air exchange rate, then the activation threshold remains unchanged;

[0108] A4. Obtain the real-time change rate of suspended particle concentration. When the suspended particle concentration deviation rate exceeds the adjusted start threshold, and the concentration change rate exceeds the preset trend threshold or the duration of exceeding the standard exceeds the preset duration, the suspended particle concentration adjustment system is triggered to start.

[0109] Input the current concentration deviation rate as the basic parameter for triggering the system. Extract historical concentration deviation rate samples of suspended particles under normal operating conditions from the historical database and calculate the set percentile of their distribution. This percentile represents the upper limit of the normal fluctuation range. When the current deviation rate exceeds this upper limit, it indicates that the current state has significantly deviated from normal. If the source tracing result is a filter abnormality or airtightness abnormality, lower the triggering threshold. If the source tracing result is insufficient ventilation, keep the threshold unchanged. On the basis of the threshold condition, add dual verification of concentration change rate or duration. Only when the deviation rate exceeds the standard and the concentration rise rate exceeds the trend threshold or the duration of exceeding the standard exceeds the preset duration will the system trigger the triggering. This helps to avoid false triggering caused by instantaneous fluctuations. When the system is running stably and the historical deviation rate distribution is concentrated, the threshold is lower and the sensitivity is higher. When the system itself fluctuates greatly, the threshold is higher to avoid frequent triggering. For high-risk faults such as filter leakage and decreased airtightness, the system automatically lowers the triggering threshold to achieve early warning. For reversible problems such as insufficient ventilation, maintain the normal threshold to avoid overreaction.

[0110] As an optional embodiment, step five specifically includes:

[0111] Incremental learning or sliding window refitting methods are used to add the initial parameters, control actions, adjustment process and final effect of this adjustment process as new samples to the historical database, and update each model periodically or triggeredly.

[0112] The complete data from each adjustment process is used as a new sample to update the model parameters online without retraining all the data. The most recent N sets of historical data, such as the last 6 months, are retained. The model is refitted periodically, and outdated data from the past is removed. Through self-updating, the model always keeps in line with the current system state. The incremental learning mechanism can quickly adapt to this change and avoid misjudgment caused by using outdated models. Continuous updates help avoid performance degradation.

[0113] like Figure 2 The intelligent industrial production environment optimization system shown includes:

[0114] The sensing module is used to collect data on suspended particle concentration, air exchange rate, filter resistance, filter integrity status, and airtightness of the maintenance structure.

[0115] The data storage module is used to store historical operating data and benchmark models;

[0116] The inference comparison module is used to compare real-time data with a benchmark model and perform anomaly tracing.

[0117] The control decision module is used to dynamically determine the start-up threshold and generate control commands;

[0118] The execution module is used to receive control commands and adjust the fan, valves and local purification unit;

[0119] The self-optimization module is used to evaluate the control effect and update the baseline model;

[0120] With clear division of labor among the modules, the system is easy to develop, test, maintain and upgrade. The perception and execution modules belong to the hardware layer, while the reasoning and decision-making modules belong to the software layer. The decoupled design allows the system to adapt to different hardware platforms. If new monitoring parameters are added in the future, only the comparison logic of the perception and reasoning modules needs to be expanded, without the need to reconstruct the overall architecture.

[0121] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A method for optimizing an intelligent industrial production environment based on historical data, characterized in that, Includes the following steps: Step 1: Construct a historical database: Record historical data of industrial production environment under different conditions to form a historical database, and establish a benchmark model based on the historical database; Step 2: Real-time data monitoring: Real-time monitoring of the current suspended particulate concentration, current air exchange rate, current resistance and integrity status of the HEPA filter, and current airtightness status of the maintenance structure in the industrial production environment to generate real-time monitoring data; Step 3, Anomaly Judgment: Compare the real-time monitoring data with the benchmark model to calculate the suspended particle concentration deviation rate, filter resistance deviation rate, and airtightness deviation rate respectively. Based on the calculation results of each deviation rate, trace the source of the anomaly, determine the dominant cause of the current suspended particle concentration anomaly, and generate the reasoning result. Step 4: Start-up of the suspended particle concentration regulation system: Based on the reasoning results, dynamically determine the start-up threshold of the suspended particle concentration regulation system, and select the corresponding control strategy to control the start-up of the suspended particle concentration regulation system. Step 5: Model self-update: After the suspended particulate concentration adjustment system is started, monitor the concentration change trend to evaluate the control effect, and feed the data of this adjustment process back to the historical database to update the baseline model.

2. The method for optimizing an intelligent industrial production environment based on historical data according to claim 1, characterized in that, Step one specifically includes: The historical database consists of historical data on suspended particle concentration in industrial production environments at different air exchange rates, historical data on the resistance and integrity tests of high-efficiency filters, and historical data on the airtightness tests of maintenance structures. The benchmark models include a benchmark relationship model between air exchange rate and suspended particle concentration, a benchmark model for filter resistance attenuation, and a benchmark model for airtightness attenuation.

3. The method for optimizing an intelligent industrial production environment based on historical data according to claim 1, characterized in that, Step three specifically includes: The method for calculating the suspended particle concentration deviation rate is as follows: in, This represents the current real-time concentration of suspended particles. The average historical normal concentration value corresponding to the current number of air exchanges is obtained from the baseline model.

4. The method for optimizing an intelligent industrial production environment based on historical data according to claim 3, characterized in that, Step three specifically includes: The filter resistance deviation rate is calculated as follows: in, The resistance of the high-efficiency filter is currently being monitored in real time. This is the expected resistance predicted based on the filter resistance attenuation benchmark model, corresponding to the current cumulative operating time of the filter.

5. The method for optimizing an intelligent industrial production environment based on historical data according to claim 4, characterized in that, Step three specifically includes: The airtightness deviation rate is calculated as follows: in, The airtightness index of the maintenance structure currently under real-time monitoring is defined as either the pressure differential decay rate or the leakage rate per unit area. It refers to the historical benchmark airtightness index obtained from the airtightness attenuation benchmark model, which corresponds to the current maintenance structure's usage time or state.

6. The method for optimizing an intelligent industrial production environment based on historical data according to claim 5, characterized in that, The specific method for anomaly tracing in step three is as follows: When the suspended particle concentration deviation rate exceeds a set first threshold and the filter resistance deviation rate exceeds a set second threshold, the primary cause is determined to be abnormal integrity of the high-efficiency filter. When the suspended particle concentration deviation rate exceeds the set first threshold and the airtightness deviation rate exceeds the set third threshold, the main cause is determined to be a decrease in the airtightness of the maintenance structure. When the suspended particle concentration deviation rate exceeds the set first threshold, while the filter resistance deviation rate and air tightness deviation rate do not exceed the corresponding thresholds, the primary cause is determined to be insufficient air exchange rate.

7. The method for optimizing an intelligent industrial production environment based on historical data according to claim 6, characterized in that, The control strategy in step four specifically includes: When the result of the anomaly tracing is that the current air exchange rate is insufficient, the variable frequency fan is controlled to increase the operating frequency and increase the air exchange rate; When the result of the anomaly tracing is that the integrity of the high-efficiency filter is abnormal, the control will issue an alarm prompt to replace or repair the high-efficiency filter, and temporarily put the backup filter unit into operation or increase the air supply volume in the adjacent area. When the result of the anomaly tracing is that the airtightness of the building envelope has decreased, the control will issue an alarm for building envelope maintenance and temporarily increase the air supply to compensate for the pollution risk caused by the leak.

8. The method for optimizing an intelligent industrial production environment based on historical data according to claim 7, characterized in that, Step four specifically includes: A1. Obtain the suspended particle concentration deviation rate by comparing the real-time monitoring data with the benchmark model; A2. Extract the historical concentration deviation rate samples of suspended particles under normal operating conditions from the historical database, calculate the set percentile of the historical deviation rate distribution of suspended particles, and use it as the start threshold. A3. Based on the anomaly tracing results, dynamically adjust the activation threshold: If the anomaly tracing result indicates an abnormality in the integrity of the high-efficiency filter, then the activation threshold is reduced to the first adjustment value; If the result of the abnormality tracing is a decrease in the airtightness of the maintenance structure, then the activation threshold is reduced to the second adjustment value; If the abnormality tracing result indicates insufficient air exchange rate, then the activation threshold remains unchanged; A4. Obtain the real-time change rate of suspended particle concentration. When the suspended particle concentration deviation rate exceeds the adjusted start threshold, and the concentration change rate exceeds the preset trend threshold or the duration of exceeding the standard exceeds the preset duration, the suspended particle concentration adjustment system is triggered to start.

9. The method for optimizing an intelligent industrial production environment based on historical data according to claim 1, characterized in that, Step five specifically includes: Incremental learning or sliding window refitting methods are used to add the initial parameters, control actions, adjustment process and final effect of this adjustment process as new samples to the historical database, and update each model periodically or triggeredly.

10. A smart industrial production environment optimization system based on historical data, applicable to the smart industrial production environment optimization method based on historical data as described in any one of claims 1-9, characterized in that, include: The sensing module is used to collect data on suspended particle concentration, air exchange rate, filter resistance, filter integrity status, and airtightness of the maintenance structure. The data storage module is used to store historical operating data and benchmark models; The inference comparison module is used to compare real-time data with a benchmark model and perform anomaly tracing. The control decision module is used to dynamically determine the start-up threshold and generate control commands; The execution module is used to receive control commands and adjust the fan, valves and local purification unit; The self-optimization module is used to evaluate the control effect and update the baseline model.

Images