Plant workshop temperature intelligent regulation method based on PID control

Abstract

The present application relates to the plant workshop temperature intelligent regulation method based on PID control, and relates to the temperature control technical field, the present application constructs digital temperature field space model and blind area compensation model, utilizes lightweight neural network to fill in the monitoring blind area between discrete sensors, realizes the real-time capture of fine temperature field under the complex flow field environment of workshop, ensures the uniformity of the regulation process, innovatively introduces the production plan and external meteorological and other multi-source feedforward interference factors, cooperates with long and short term memory network, has the predictive scheduling function, effectively deals with the large lag problem caused by large space thermal inertia, and dynamically adjusts the PID parameters based on the prediction trend, realizes the on-demand accurate energy supply of each partition, and the output smooth variable frequency control instruction avoids the shock and energy waste caused by the frequent start of the actuator, greatly reduces the comprehensive operation cost of the plant.

Description

Technical Field

[0001] This invention relates to the field of temperature control technology, and in particular to a method for intelligent temperature control in factory workshops based on PID control. Background Technology

[0002] In the current context of Industry 4.0 and green manufacturing, although intelligent temperature control in factories and workshops has evolved from traditional manual switching to automated control, existing mainstream methods still have the following core shortcomings when facing complex industrial environments:

[0003] I. Spatial Blind Spots and Lag in Sensing: Most factories use a discrete sensor layout, which limits sensing capabilities. Sensors are typically mounted on walls or fixed supports, failing to reflect the microenvironment in the center of the workshop, the ceiling of tall spaces, or around heat-generating equipment in real time. This point-to-area approach makes the control system unable to cope with localized overheating or uneven heating. Furthermore, existing low-power sensors exhibit significant hysteresis when the environment changes drastically, meaning that by the time the system triggers control, the ambient temperature may have already deviated from the preset range.

[0004] Second, the generalization ability of the model algorithm is insufficient: Current intelligent control is mostly based on fixed PID control or simple expert rule bases. However, workshop temperature is affected by a variety of nonlinear factors such as equipment heat dissipation, door and window opening and closing, outdoor weather, and personnel flow. Existing models often cannot effectively handle these multivariate couplings, leading to control oscillations. Most systems are reactive, meaning they only start cooling / heating after the temperature exceeds the limit. Due to the extremely high thermal inertia of large spaces, this passive adjustment leads to a surge in energy consumption and large environmental fluctuations. To address the aforementioned technical deficiencies, a solution is proposed. Summary of the Invention

[0005] The purpose of this invention is to: construct a digital temperature field spatial model and a blind zone compensation model, and use a lightweight neural network to fill the monitoring blind zone between discrete sensors. This overcomes the perception limitations caused by the sparse sensor deployment in traditional methods, enabling real-time capture of a refined temperature field in a complex flow environment in a workshop. This ensures the uniformity of the control process. It innovatively introduces multi-source feedforward interference factors such as production plans and external weather conditions, and, in conjunction with a long short-term memory network, provides predictive scheduling capabilities. This effectively addresses the large lag problem caused by thermal inertia in large spaces, and dynamically adjusts PID parameters based on predicted trends to achieve precise on-demand energy supply to each zone. The output smooth frequency conversion control commands avoid oscillations and energy waste caused by frequent start-ups of actuators, significantly reducing the overall operating costs of the plant.

[0006] 2. To achieve the above objectives, the present invention adopts the following technical solution: a method for intelligent temperature control in factory workshops based on PID control, characterized by comprising the following steps: S1. Construct a digital temperature field spatial model: Obtain building structure data and internal equipment distribution data of the factory workshop to establish a three-dimensional spatial grid model, and mark the discrete sensors set in the factory workshop in the three-dimensional spatial grid model as detection node coordinates; S2. Constructing a blind zone compensation model: Obtain real-time monitoring data from each discrete sensor, divide the effective monitoring area and monitoring blind zone within the three-dimensional spatial grid model according to the coordinates of the detection nodes, and use computational fluid dynamics pre-simulation data to train a lightweight neural network to construct a blind zone compensation model, thereby realizing real-time interpolation compensation of the temperature in the monitoring blind zone between the sampling points of discrete sensors and obtaining refined temperature field data. S3, Multi-source interference factor perception: By connecting with the production management system, the operating logic of equipment in the factory workshop is obtained, and real-time meteorological parameters are obtained by external meteorological sensors, which are integrated into the interference factor as the interference feedforward input of the PID control system. S4. Calculation of predictive PID control quantity: Input the refined temperature field data obtained in S2 and the disturbance factor obtained in S3 into the pre-built predictive control model, use the long short-term memory network to predict the temperature change trend within the future preset period, and calculate the control compensation quantity. S5. Dynamic adjustment execution: Based on the analysis and prediction of the trend of the control compensation quantity, the proportional, integral and derivative parameters of the PID controller are adjusted, and the frequency conversion control command is output to the terminal air conditioning and ventilation equipment in each zone of the factory workshop to achieve temperature regulation.

[0007] 2. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that the specific process of establishing a three-dimensional spatial grid model is as follows: S101. Use a LiDAR scanner to acquire building structure data and internal equipment distribution data of the factory workshop. The building structure data includes the length, width, and height of the workshop, the physical location and opening direction of the doors and windows, and the distribution data of load-bearing columns and partition walls. The internal equipment distribution data includes the center coordinates, external dimensions, and heat dissipation power characteristics of the heat-generating equipment. S102. Construct a three-dimensional basic model based on building structure data and internal equipment distribution data, obtain the three-dimensional spatial coordinates of discrete sensors deployed in the factory workshop, and embed the three-dimensional spatial coordinates as known constraint points into the three-dimensional basic model. S103. Non-uniform gridding technology is used to discretize the physical space of the three-dimensional basic model. A dense grid is used around the heating equipment, the air conditioning vent area, and the door and window boundaries. A sparse grid is used in areas with relatively stable temperature fields. S104. Define each grid center after division as a computing node. Each computing node is assigned a set of physical attributes, which includes the initial temperature value, air velocity, and the partition identifier to which it belongs. S105. Take a corner point of the factory workshop entrance ground as the origin and establish a standard right-hand rectangular coordinate system. Map the coordinates of discrete sensors to the corresponding grid nodes. If the physical location of the discrete sensor is not in the center of the grid, use the nearest neighbor mapping algorithm or the linear weighted algorithm to associate the real-time monitoring data of the discrete sensor with the N neighboring grid nodes as the initial boundary conditions of the blind zone compensation model. S106. Set the grid surface corresponding to the outer wall of the factory workshop as the heat exchange boundary. The heat transfer coefficient of the heat exchange boundary is dynamically adjusted with the external meteorological parameters. Define the grid area corresponding to the production equipment as the volume heat source item. The heating intensity of the volume heat source item is dynamically assigned according to the equipment load percentage obtained by the production management system.

[0008] 3. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that the specific process of obtaining refined temperature field data is as follows: S201. Based on the three-dimensional spatial grid model constructed in S1, a spherical space with a radius of R is set as the effective monitoring area, with the three-dimensional spatial coordinates of each discrete sensor as the center. The temperature field data in the effective monitoring area is directly taken from the real-time monitoring value of the sensor and attenuated towards the edge through a Gaussian weighted algorithm. S202. The grid nodes in the workshop that do not belong to any effective monitoring area are defined as monitoring blind zones. The monitoring blind zones are located where large equipment is obstructed, at the top of tall spaces, and in the center of the workshop away from the walls. S203. Simulate multiple combinations of variables to generate multiple sets of global grid temperature distribution data. Using the three-dimensional spatial coordinates in the simulation as input and the corresponding global grid temperature data as labels, train a lightweight convolutional neural network to obtain the blind spot compensation model. Use knowledge distillation technology to ensure that the blind spot compensation model can achieve millisecond-level real-time inference in the edge computing gateway and obtain the pre-simulation baseline value. S204. Acquire real-time monitoring data from each discrete sensor and inject the real-time monitoring data into the trained blind zone compensation model. By extracting the spatial correlation features between detection nodes, identify the current heat convection pattern in the workshop. S205. Based on the deviation between real-time monitoring data and pre-simulation benchmark values, perform nonlinear interpolation calculations on the grid nodes in the monitoring blind zone. S206. The real-time monitoring data of the effective monitoring area is fused with the pre-simulation benchmark value of the monitoring blind area. Laplace smoothing is used to eliminate temperature jumps at the boundary of the monitoring area, ensuring the continuity of spatial temperature distribution. A three-dimensional temperature matrix containing the coordinates of all grid nodes in the workshop and their corresponding temperature values ​​is output. The three-dimensional temperature matrix is ​​the refined temperature field data.

[0009] 4. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that the specific process of obtaining the interference factor is as follows: S301. Real-time data interaction with the production management system of the workshop through a standard communication interface to obtain equipment operation logic. The equipment operation logic includes production work order scheduling for the next N hours, so as to extract the preheating time, operating load and shutdown time of each production line's heat-generating equipment, convert the operating load of each piece of equipment into equivalent heat generation, and obtain the equipment interference factor. S302. Calculate the degree of personnel gathering at each workstation in the workshop based on the shift schedule information, and convert it into a human body heat dissipation interference factor per unit area. S303. Outdoor environmental parameters are obtained by deploying a high-precision integrated meteorological station outside the workshop and factory buildings, and real-time outdoor temperature and solar radiation intensity are obtained. The heat received by the factory roof and exterior walls is calculated based on the solar radiation intensity. The heat transfer disturbance of the heat received on the internal edge grid nodes is calculated by the heat transfer coefficient equation to obtain the environmental disturbance factor. S304. Acquire wind direction data, wind speed data and outdoor humidity data. The wind speed data is used to evaluate the natural convection exchange when workshop doors and windows and ventilation fans are turned on, and is integrated into climate disturbance factors. S305. Align the timestamps of equipment interference factor, human body heat dissipation interference factor, environmental interference factor and climate interference factor to the current sampling time, and accurately map the spatial interference sources to the three-dimensional spatial grid model constructed in S1. Use the correlation analysis algorithm to calculate the contribution rate of each interference factor to the temperature field fluctuation. S306. The quantized interference factors described above are integrated into an interference feedforward feature vector. : , where H out Represents the influence of external heat exchange. P represents the change in the internal heat source. S Represents production scheduling forecasting, W E This represents meteorological and dynamic disturbances.

[0010] 5. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that the specific process for calculating the control compensation amount is as follows: S401. Extract historical data from a preset preceding period to the current time t, and construct a time-series feature matrix in the form of a sliding window. The time-series feature matrix includes: Spatial dimension input: Obtain refined temperature field data from S2 output, and extract the average temperature and temperature gradient of key zones; Interference dimension input: Obtain the integrated interference feedforward feature vector F in S3, including the predicted equipment heat generation fluctuation and outdoor weather evolution within the preset period; Status dimension input: Obtain the operating frequency and on / off status of the currently running temperature control equipment; S402. Input the time-series feature matrix into the pre-trained predictive control model. The predictive control model, through forget gate and memory gate structures, can capture the long-range time dependence caused by the thermal inertia of the large space in the workshop, so as to output the temperature evolution curve within the future preset period. The temperature evolution curve reflects the natural temperature rise and temperature drop trend after being affected by the current production plan and the external environment without additional human intervention. S403. Compare the temperature evolution curve with the preset target temperature and calculate the feedforward compensation logic: Prediction bias is defined based on the thermodynamic equilibrium equation. The formula is expressed as follows: Where t is the current time, Pre-set a cycle for the future. The target temperature is set to T, where T is the temperature value within the temperature evolution curve. S404. Obtain the workshop space volume and air density provided by the three-dimensional spatial grid model, calculate the heat exchange amount required to offset the prediction deviation, and convert the heat exchange into the control compensation amount of the control domain.

[0011] 6. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that the specific process of outputting the frequency conversion control command is as follows: S501: Obtain the control compensation quantity and prediction deviation calculated in S4, and use fuzzy inference to correct the three core parameters of the traditional PID controller in real time. When the prediction deviation exceeds the preset deviation threshold, the PID controller's response speed to future heat load changes is improved based on the proportional coefficient of the increase in prediction deviation. Based on the static deviation of each region in the refined temperature field data, the integral coefficient is dynamically fine-tuned to eliminate the residual static error caused by thermal hysteresis in large spaces. By combining the slope of change brought about by each interference factor, the weight of the differential coefficient is increased, and the anticipatory adjustment characteristic of the differential term is used to preemptively offset the impact caused by the instantaneous high-power start-up and shutdown of the equipment. S502. Decompose the total control quantity of the PID controller according to the partition attributes in the three-dimensional spatial grid model: High heat load area: Allocate a higher power gain coefficient to terminal air conditioning equipment located around heat-generating equipment; Stable temperature zone: Energy-saving operation mode is adopted for areas far from the center of the heat source, maintaining low-frequency operation; Boundary coupling zone: Combined with the heat transfer interference from the external wall, the speed of the air curtain machine or ventilator near the doors and windows is dynamically adjusted to form a thermal air shielding zone; S503. Encode the calculated control quantities of each zone into a standard industrial control protocol, and map each zone control quantity to a specific operating frequency according to the rated power range of the actuator.

[0012] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: This intelligent temperature control method for factory workshops based on PID control constructs a digital temperature field spatial model and a blind zone compensation model. It utilizes a lightweight neural network to fill the monitoring blind zones between discrete sensors, overcoming the perception limitations caused by the sparse sensor deployment in traditional methods. This method achieves real-time capture of a refined temperature field in the complex flow field environment of the workshop, ensuring the uniformity of the control process. It innovatively introduces multi-source feedforward interference factors such as production plans and external meteorological conditions. Combined with a long short-term memory network, it has predictive scheduling capabilities, effectively addressing the large lag problem caused by thermal inertia in large spaces. Based on predicted trends, it dynamically adjusts PID parameters to achieve precise on-demand energy supply to each zone. The output smooth frequency conversion control commands avoid oscillations and energy waste caused by frequent start-ups of actuators, significantly reducing the overall operating costs of the factory. Attached Figure Description

[0013] Figure 1 A schematic diagram of the overall method flow of the present invention is shown. Detailed Implementation

[0014] 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.

[0015] Example: like Figure 1 As shown, the intelligent temperature control method for factory workshops based on PID control includes the following steps: S1. Construct a digital temperature field spatial model: Obtain building structure data and internal equipment distribution data of the factory workshop to establish a three-dimensional spatial grid model, and mark the discrete sensors set in the factory workshop in the three-dimensional spatial grid model as detection node coordinates; The specific process of establishing a three-dimensional spatial grid model is as follows: S101. Use a LiDAR scanner to acquire building structure data and internal equipment distribution data of the factory workshop. The building structure data includes the length, width, and height of the workshop, the physical location and opening direction of the doors and windows, and the distribution data of load-bearing columns and partition walls. The internal equipment distribution data includes the center coordinates, external dimensions, and heat dissipation power characteristics of heat-generating equipment (such as injection molding machines, ovens, and large machine tools). S102. Construct a three-dimensional basic model based on building structure data and internal equipment distribution data, obtain the three-dimensional spatial coordinates of discrete sensors deployed in the factory workshop, and embed the three-dimensional spatial coordinates as known constraint points into the three-dimensional basic model. S103. Non-uniform gridding technology is used to discretize the physical space of the three-dimensional basic model. A dense grid is used around the heating equipment, the air conditioning vent area, and the door and window boundaries. A sparse grid is used in areas with relatively stable temperature fields. S104. Define each grid center after division as a computing node. Each computing node is assigned a set of physical attributes, which includes the initial temperature value, air velocity, and the partition identifier to which it belongs. S105. Take a corner point of the factory workshop entrance ground as the origin and establish a standard right-hand rectangular coordinate system. Map the coordinates of discrete sensors to the corresponding grid nodes. If the physical location of the discrete sensor is not in the center of the grid, use the nearest neighbor mapping algorithm or the linear weighted algorithm to associate the real-time monitoring data of the discrete sensor with the N neighboring grid nodes as the initial boundary conditions of the blind zone compensation model. S106. Set the grid surface corresponding to the outer wall of the factory workshop as the heat exchange boundary. The heat transfer coefficient of the heat exchange boundary is dynamically adjusted with the external meteorological parameters. Define the grid area corresponding to the production equipment as the volume heat source item. The heating intensity of the volume heat source item is dynamically assigned according to the equipment load percentage obtained by the production management system.

[0016] S2. Constructing a blind zone compensation model: Obtain real-time monitoring data from each discrete sensor, divide the effective monitoring area and monitoring blind zone within the three-dimensional spatial grid model according to the coordinates of the detection nodes, and use computational fluid dynamics pre-simulation data to train a lightweight neural network to construct a blind zone compensation model, thereby realizing real-time interpolation compensation of the temperature in the monitoring blind zone between the sampling points of discrete sensors and obtaining refined temperature field data. The specific process for obtaining refined temperature field data is as follows: S201. Based on the three-dimensional spatial grid model constructed in S1, a spherical space with a radius of R is set as the effective monitoring area, with the three-dimensional spatial coordinates of each discrete sensor as the center. The temperature field data in the effective monitoring area is directly taken from the real-time monitoring value of the sensor and attenuated towards the edge through a Gaussian weighted algorithm. S202. The grid nodes in the workshop that do not belong to any effective monitoring area are defined as monitoring blind zones. The monitoring blind zones are located where large equipment is obstructed, at the top of tall spaces, and in the center of the workshop away from the walls. S203. Simulate multiple combinations of variables to generate multiple sets of global grid temperature distribution data. Using the three-dimensional spatial coordinates in the simulation as input and the corresponding global grid temperature data as labels, train a lightweight convolutional neural network to obtain the blind spot compensation model. Use knowledge distillation technology to ensure that the blind spot compensation model can achieve millisecond-level real-time inference in the edge computing gateway and obtain the pre-simulation baseline value. S204. Acquire real-time monitoring data from each discrete sensor and inject the real-time monitoring data into the trained blind zone compensation model. By extracting the spatial correlation features between detection nodes, identify the current heat convection pattern in the workshop. S205. Based on the deviation between real-time monitoring data and pre-simulation benchmark values, perform nonlinear interpolation calculations on the grid nodes in the monitoring blind zone. For example, when sensors A and B are heated simultaneously, the model automatically calculates the temperature evolution law of the blind zone nodes between them based on the fluid topology. S206. The real-time monitoring data of the effective monitoring area is fused with the pre-simulation benchmark value of the monitoring blind area. Laplace smoothing is used to eliminate temperature jumps at the boundary of the monitoring area, ensuring the continuity of spatial temperature distribution. A three-dimensional temperature matrix containing the coordinates of all grid nodes in the workshop and their corresponding temperature values ​​is output. The three-dimensional temperature matrix is ​​the refined temperature field data, and its spatial resolution is much higher than that of the physical sensor array.

[0017] S3, Multi-source interference factor perception: By connecting with the production management system, the operating logic of equipment in the factory workshop is obtained, and real-time meteorological parameters are obtained by external meteorological sensors, which are integrated into the interference factor as the interference feedforward input of the PID control system. The specific process for obtaining the interference factor is as follows: S301. Real-time data interaction with the production management system of the workshop through a standard communication interface to obtain equipment operation logic. The equipment operation logic includes production work order scheduling for the next N hours, to extract the preheating time, operating load and shutdown time of each production line's heat-generating equipment, convert the operating load of each piece of equipment into equivalent heat generation, and obtain the equipment interference factor. For example, according to the pre-stored equipment power consumption model, the 80% load operation state of the injection molding machine is mapped to the heat source intensity item of the corresponding coordinate point in the grid model. S302. Calculate the degree of personnel gathering at each workstation in the workshop based on the shift schedule information, and convert it into a human body heat dissipation interference factor per unit area. S303. Outdoor environmental parameters are obtained by deploying a high-precision integrated meteorological station outside the workshop and factory buildings, and real-time outdoor temperature and solar radiation intensity are obtained. The heat received by the factory roof and exterior walls is calculated based on the solar radiation intensity. The heat transfer disturbance of the heat received on the internal edge grid nodes is calculated by the heat transfer coefficient equation to obtain the environmental disturbance factor. S304. Acquire wind direction data, wind speed data and outdoor humidity data. The wind speed data is used to evaluate the natural convection exchange when the workshop doors, windows and ventilation fans are turned on, and is integrated into a climate interference factor, which directly affects the temperature fluctuation in the edge area of ​​the workshop. S305. Align the timestamps of equipment interference factors, human body heat dissipation interference factors, environmental interference factors, and climate interference factors to the current sampling time, and accurately map the spatial interference sources into the three-dimensional spatial grid model constructed in S1. Use a correlation analysis algorithm to calculate the contribution rate of each interference factor to temperature field fluctuations. For example, in light steel workshops with poor thermal insulation performance, increase the weight of the "solar radiation" factor; in precision machining workshops, increase the weight of the "equipment start-up and shutdown" factor. S306. The quantized interference factors described above are integrated into an interference feedforward feature vector. : , where H out Represents the influence of external heat exchange. P represents the change in the internal heat source. S Represents production scheduling forecasting, W E This represents meteorological and dynamic disturbances.

[0018] S4. Calculation of predictive PID control quantity: Input the refined temperature field data obtained in S2 and the disturbance factor obtained in S3 into the pre-built predictive control model, use the long short-term memory network to predict the temperature change trend within the future preset period, and calculate the control compensation quantity. The specific process for calculating the control compensation amount is as follows: S401. Extract historical data from a preset preceding period to the current time t, and construct a time-series feature matrix in the form of a sliding window. The time-series feature matrix includes: Spatial dimension input: Obtain refined temperature field data from S2 output, and extract the average temperature and temperature gradient of key zones (such as high-precision processing areas and densely populated areas); Interference dimension input: Obtain the integrated interference feedforward feature vector F in S3, including the predicted equipment heat generation fluctuation and outdoor weather evolution within the preset period; Status dimension input: Obtain the operating frequency and on / off status of currently running temperature control equipment (air conditioner, fan); S402. Input the time-series feature matrix into the pre-trained predictive control model. The predictive control model, through forget gate and memory gate structures, can capture the long-range time dependence caused by the thermal inertia of the large space in the workshop, so as to output the temperature evolution curve within the future preset period. The temperature evolution curve reflects the natural temperature rise and temperature drop trend after being affected by the current production plan and the external environment without additional human intervention. S403. Compare the temperature evolution curve with the preset target temperature and calculate the feedforward compensation logic: Prediction bias is defined based on the thermodynamic equilibrium equation. The formula is expressed as follows: Where t is the current time, Pre-set a cycle for the future. The target temperature is set to T, where T is the temperature value within the temperature evolution curve. S404. Obtain the workshop space volume and air density provided by the three-dimensional spatial grid model, calculate the heat exchange amount required to offset the prediction deviation, and convert the heat exchange into the control compensation amount of the control domain. The control compensation amount is used as a feedforward term to correct the adjustment lag caused by pure feedback in traditional PID control. To prevent drastic oscillations in control commands, the calculated control compensation values ​​are smoothed. A first-order hysteresis filtering algorithm is adopted to ensure that the intervention process of the control compensation is smooth, avoid frequent and drastic speed adjustment of the frequency converter leading to equipment damage, and output the final effective control compensation to step S5.

[0019] S5. Dynamic adjustment execution: Based on the analysis and prediction of the trend of the control compensation quantity, the proportional, integral and derivative parameters of the PID controller are adjusted, and the frequency conversion control command is output to the terminal air conditioning and ventilation equipment in each zone of the factory workshop to achieve temperature regulation.

[0020] The specific process of outputting frequency converter control commands is as follows: S501: Obtain the control compensation quantity and prediction deviation calculated in S4, and use fuzzy inference to correct the three core parameters of the traditional PID controller in real time. When the prediction deviation exceeds the preset deviation threshold, the PID controller's response speed to future heat load changes is improved based on the proportional coefficient of the increase in prediction deviation. Based on the static deviation of each region in the refined temperature field data, the integral coefficient is dynamically fine-tuned to eliminate the residual static error caused by thermal hysteresis in large spaces and prevent the occurrence of control dead zones when the target temperature is reached. By combining the slope of change brought about by each interference factor, the weight of the differential coefficient is increased, and the anticipatory adjustment characteristic of the differential term is used to preemptively offset the impact caused by the instantaneous high-power start-up and shutdown of the equipment. S502. Decompose the total control quantity of the PID controller according to the partition attributes in the three-dimensional spatial grid model: High heat load area: Allocate a higher power gain coefficient to terminal air conditioning equipment located around heat-generating equipment; Stable temperature zone: Energy-saving operation mode is adopted for areas far from the center of the heat source, maintaining low-frequency operation; Boundary coupling zone: Combined with the heat transfer interference from the external wall, the speed of the air curtain machine or ventilator near the doors and windows is dynamically adjusted to form a thermal air shielding zone; S503. Encode the calculated control quantities of each zone into a standard industrial control protocol, and map each zone control quantity to a specific operating frequency according to the rated power range of the actuator (such as the host frequency of the water system air conditioner or the speed of the fan in the air system). To prevent motor oscillation caused by frequent adjustments to PID parameters, the system has a built-in ramp function, which sends frequency conversion commands to the frequency converter in a smooth curve, avoiding instantaneous load fluctuations in the power grid and extending the service life of the equipment.

[0021] Simultaneously, after outputting the frequency conversion command, the feedback data of the refined temperature field in S2 is continuously monitored: If the actual temperature rise trajectory deviates from the predicted curve, the deviation is fed back to the predictive control model in S4 for online self-correction, and the PID gain strategy in S5 is adjusted simultaneously to form a complete closed loop of perception-prediction-execution-feedback.

[0022] This invention constructs a digital temperature field spatial model and a blind zone compensation model, and uses a lightweight neural network to fill the monitoring blind zone between discrete sensors. It overcomes the perception limitations caused by the sparse sensor deployment in traditional methods, and achieves real-time capture of a refined temperature field in the complex flow field environment of the workshop, ensuring the uniformity of the control process. It innovatively introduces multi-source feedforward interference factors such as production planning and external meteorological conditions, and, in conjunction with a long short-term memory network, has predictive scheduling capabilities, effectively addressing the large lag problem caused by thermal inertia in large spaces. Based on the predicted trend, it dynamically adjusts PID parameters to achieve precise on-demand energy supply to each zone. The output smooth frequency conversion control commands avoid oscillations and energy waste caused by frequent start-ups of actuators, significantly reducing the overall operating cost of the plant.

[0023] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value.

[0024] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent temperature control in factory workshops based on PID control, characterized in that, Includes the following steps: S1. Construct a digital temperature field spatial model: Obtain building structure data and internal equipment distribution data of the factory workshop to establish a three-dimensional spatial grid model, and mark the discrete sensors set in the factory workshop in the three-dimensional spatial grid model as detection node coordinates; S2. Constructing a blind zone compensation model: Obtain real-time monitoring data from each discrete sensor, divide the effective monitoring area and monitoring blind zone within the three-dimensional spatial grid model according to the coordinates of the detection nodes, and use computational fluid dynamics pre-simulation data to train a lightweight neural network to construct a blind zone compensation model, thereby realizing real-time interpolation compensation of the temperature in the monitoring blind zone between the sampling points of discrete sensors and obtaining refined temperature field data. S3, Multi-source interference factor perception: By connecting with the production management system, the operating logic of equipment in the factory workshop is obtained, and real-time meteorological parameters are obtained by external meteorological sensors, which are integrated into the interference factor as the interference feedforward input of the PID control system. S4. Calculation of predictive PID control quantity: Input the refined temperature field data obtained in S2 and the disturbance factor obtained in S3 into the pre-built predictive control model, use the long short-term memory network to predict the temperature change trend within the future preset period, and calculate the control compensation quantity. S5. Dynamic adjustment execution: Based on the analysis and prediction of the trend of the control compensation quantity, the proportional, integral and derivative parameters of the PID controller are adjusted, and the frequency conversion control command is output to the terminal air conditioning and ventilation equipment in each zone of the factory workshop to achieve temperature regulation.

2. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that, The specific process of establishing a three-dimensional spatial grid model is as follows: S101. Use a LiDAR scanner to acquire building structure data and internal equipment distribution data of the factory workshop. The building structure data includes the length, width, and height of the workshop, the physical location and opening direction of the doors and windows, and the distribution data of load-bearing columns and partition walls. The internal equipment distribution data includes the center coordinates, external dimensions, and heat dissipation power characteristics of the heat-generating equipment. S102. Construct a three-dimensional basic model based on building structure data and internal equipment distribution data, obtain the three-dimensional spatial coordinates of discrete sensors deployed in the factory workshop, and embed the three-dimensional spatial coordinates as known constraint points into the three-dimensional basic model. S103. Non-uniform gridding technology is used to discretize the physical space of the three-dimensional basic model. A dense grid is used around the heating equipment, the air conditioning vent area, and the door and window boundaries. A sparse grid is used in areas with relatively stable temperature fields. S104. Define each grid center after division as a computing node. Each computing node is assigned a set of physical attributes, which includes the initial temperature value, air velocity, and the partition identifier to which it belongs. S105. Take a corner point of the factory workshop entrance ground as the origin and establish a standard right-hand rectangular coordinate system. Map the coordinates of discrete sensors to the corresponding grid nodes. If the physical location of the discrete sensor is not in the center of the grid, use the nearest neighbor mapping algorithm or the linear weighted algorithm to associate the real-time monitoring data of the discrete sensor with the N neighboring grid nodes as the initial boundary conditions of the blind zone compensation model. S106. Set the grid surface corresponding to the outer wall of the factory workshop as the heat exchange boundary. The heat transfer coefficient of the heat exchange boundary is dynamically adjusted with the external meteorological parameters. Define the grid area corresponding to the production equipment as the volume heat source item. The heating intensity of the volume heat source item is dynamically assigned according to the equipment load percentage obtained by the production management system.

3. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that, The specific process for obtaining refined temperature field data is as follows: S201. Based on the three-dimensional spatial grid model constructed in S1, a spherical space with a radius of R is set as the effective monitoring area, with the three-dimensional spatial coordinates of each discrete sensor as the center. The temperature field data in the effective monitoring area is directly taken from the real-time monitoring value of the sensor and attenuated towards the edge through a Gaussian weighted algorithm. S202. The grid nodes in the workshop that do not belong to any effective monitoring area are defined as monitoring blind zones. The monitoring blind zones are located where large equipment is obstructed, at the top of tall spaces, and in the center of the workshop away from the walls. S203. Simulate multiple combinations of variables to generate multiple sets of global grid temperature distribution data. Using the three-dimensional spatial coordinates in the simulation as input and the corresponding global grid temperature data as labels, train a lightweight convolutional neural network to obtain the blind spot compensation model. Use knowledge distillation technology to ensure that the blind spot compensation model can achieve millisecond-level real-time inference in the edge computing gateway and obtain the pre-simulation baseline value. S204. Acquire real-time monitoring data from each discrete sensor and inject the real-time monitoring data into the trained blind zone compensation model. By extracting the spatial correlation features between detection nodes, identify the current heat convection pattern in the workshop. S205. Based on the deviation between real-time monitoring data and pre-simulation benchmark values, perform nonlinear interpolation calculations on the grid nodes in the monitoring blind zone. S206. The real-time monitoring data of the effective monitoring area is fused with the pre-simulation benchmark value of the monitoring blind area. Laplace smoothing is used to eliminate temperature jumps at the boundary of the monitoring area, ensuring the continuity of spatial temperature distribution. A three-dimensional temperature matrix containing the coordinates of all grid nodes in the workshop and their corresponding temperature values ​​is output. The three-dimensional temperature matrix is ​​the refined temperature field data.

4. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that, The specific process for obtaining the interference factor is as follows: S301. Real-time data interaction with the production management system of the workshop through a standard communication interface to obtain equipment operation logic. The equipment operation logic includes production work order scheduling for the next N hours, so as to extract the preheating time, operating load and shutdown time of each production line's heat-generating equipment, convert the operating load of each piece of equipment into equivalent heat generation, and obtain the equipment interference factor. S302. Calculate the degree of personnel gathering at each workstation in the workshop based on the shift schedule information, and convert it into a human body heat dissipation interference factor per unit area. S303. Outdoor environmental parameters are obtained by deploying a high-precision integrated meteorological station outside the workshop and factory buildings, and real-time outdoor temperature and solar radiation intensity are obtained. The heat received by the factory roof and exterior walls is calculated based on the solar radiation intensity. The heat transfer disturbance of the heat received on the internal edge grid nodes is calculated by the heat transfer coefficient equation to obtain the environmental disturbance factor. S304. Acquire wind direction data, wind speed data and outdoor humidity data. The wind speed data is used to evaluate the natural convection exchange when workshop doors and windows and ventilation fans are turned on, and is integrated into climate disturbance factors. S305. Align the timestamps of equipment interference factor, human body heat dissipation interference factor, environmental interference factor and climate interference factor to the current sampling time, and accurately map the spatial interference sources to the three-dimensional spatial grid model constructed in S1. Use the correlation analysis algorithm to calculate the contribution rate of each interference factor to the temperature field fluctuation. S306. The quantized interference factors described above are integrated into an interference feedforward feature vector. : , where H out Represents the influence of external heat exchange. P represents changes in the internal heat source. S Represents production scheduling forecasting, W E This represents meteorological and dynamic interference.

5. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that, The specific process for calculating the control compensation amount is as follows: S401. Extract historical data from a preset preceding period to the current time t, and construct a time-series feature matrix in the form of a sliding window. The time-series feature matrix includes: Spatial dimension input: Obtain refined temperature field data from S2 output, and extract the average temperature and temperature gradient of key zones; Interference dimension input: Obtain the integrated interference feedforward feature vector F in S3, including the predicted equipment heat generation fluctuation and outdoor weather evolution within the preset period; Status dimension input: Obtain the operating frequency and on / off status of the currently running temperature control equipment; S402. Input the time-series feature matrix into the pre-trained predictive control model. The predictive control model, through forget gate and memory gate structures, can capture the long-range time dependence caused by the thermal inertia of the large space in the workshop, so as to output the temperature evolution curve within the future preset period. The temperature evolution curve reflects the natural temperature rise and temperature drop trend after being affected by the current production plan and the external environment without additional human intervention. S403. Compare the temperature evolution curve with the preset target temperature and calculate the feedforward compensation logic: Prediction bias is defined based on the thermodynamic equilibrium equation. The formula is expressed as follows: Where t is the current time, Pre-set a cycle for the future. The target temperature is set to T, where T is the temperature value within the temperature evolution curve. S404. Obtain the workshop space volume and air density provided by the three-dimensional spatial grid model, calculate the heat exchange amount required to offset the prediction deviation, and convert the heat exchange into the control compensation amount of the control domain.

6. The intelligent temperature control method for factory workshops based on PID control according to claim 1, characterized in that, The specific process of outputting frequency converter control commands is as follows: S501: Obtain the control compensation quantity and prediction deviation calculated in S4, and use fuzzy inference to correct the three core parameters of the traditional PID controller in real time. When the prediction deviation exceeds the preset deviation threshold, the PID controller's response speed to future heat load changes is improved based on the proportional coefficient of the increase in prediction deviation. Based on the static deviation of each region in the refined temperature field data, the integral coefficient is dynamically fine-tuned to eliminate the residual static error caused by thermal hysteresis in large spaces. By combining the slope of change brought about by each interference factor, the weight of the differential coefficient is increased, and the anticipatory adjustment characteristic of the differential term is used to preemptively offset the impact caused by the instantaneous high-power start-up and shutdown of the equipment. S502. Decompose the total control quantity of the PID controller according to the partition attributes in the three-dimensional spatial grid model: High heat load area: Allocate a higher power gain coefficient to terminal air conditioning equipment located around heat-generating equipment; Stable temperature zone: Energy-saving operation mode is adopted for areas far from the center of the heat source, maintaining low-frequency operation; Boundary coupling zone: Combined with the heat transfer interference from the external wall, the speed of the air curtain machine or ventilator near the doors and windows is dynamically adjusted to form a thermal air shielding zone; S503. Encode the calculated control quantities of each zone into a standard industrial control protocol, and map each zone control quantity to a specific operating frequency according to the rated power range of the actuator.

Images