JP7531258B2 - Air Conditioning Control System - Google Patents

Description

Article 30, paragraph 2 of the Patent Act applies ▲1▼ Publication date April 5, 2020 Publication Clean Technology April 2020 issue Pages 27-30 Japan Industrial Publishing Co., Ltd. ▲2▼ Publication date April 21, 2020 Publication Proceedings of the Air Purification and Contamination Control Research Conference Pages 45-48 C-3 Japan Air Cleaning Association Publication

The present invention relates to an air conditioning control system that realizes both optimized temperature control and energy conservation, and in particular to an air conditioning control system for industrial clean rooms used in semiconductor factories, factories that manufacture films, liquid crystals, and organic electroluminescence (EL) films used in flat panel displays, precision machinery factories, etc.

In industrial clean room air conditioning, even small amounts of airborne particles can adhere to products being manufactured and cause product defects, so work rooms are kept clean as needed to improve product quality and reliability and increase yields.

Recently, the idea of mini-environments and micro-environments, which locally purify only the areas where semi-finished products or finished products are manufactured or transported between processes, has become widespread. In reality, this is a technical idea that isolates the space where semi-finished products or finished products are exposed from the rest as a small space and maintains a locally high level of cleanliness within this small space, and has been introduced into production equipment, transport between processes, etc.

With this method, the areas within the clean room where particularly high levels of cleanliness are required are limited to certain areas, and other areas do not require the same high levels of cleanliness as in the past, so the room as a whole only needs to maintain a certain level of cleanliness. As a result, large spaces (ballrooms) have become the mainstream for industrial clean rooms.

Even in these large-area (ballroom)-style clean rooms, the air flow is generally from top to bottom, although it differs from the old full-area downflow that required a lot of electricity. A ceiling outlet with a HEPA filter at the end is installed in part of the ceiling, and air is drawn into the space between the slab floor and the raised floor (underfloor chamber) through the openings in the raised floor, which is formed by perforated and non-perforated panels with small holes in the floor, and the air is then temperature-regulated and blown out again from the ceiling to remove dust. This is how dust generated inside the room is quickly eliminated.

Figure 4 is a schematic diagram illustrating an example of a conventional clean room. As shown in Figure 4, the conventional clean room is installed in a clean room building 20 that is relatively airtight, such as a semiconductor factory, and is composed of a ceiling supply chamber 14 formed between the ceiling 21 of the building 20 and the ceiling 15 of the work area (clean room) 11, in which a lattice is formed using T-bars or channel materials, an underfloor chamber 17 formed between the floor surface 16 of the clean room (hereinafter also referred to as the work area) 11 and the floor slab 23 of the clean room building 20, a return shaft formed between the airtight side wall 22 of the clean room building 20 and the partition wall 12 of the work area 11, which is formed using boards or panels, a fan filter unit (FFU) 2 installed on the ceiling 15 of the work area 11, and an air cooling means (dry coil) 3 for cooling the return air.

The air in the ceiling supply chamber 14 is then cleaned by the fan filter unit 2 to ensure a specified level of cleanliness, and is then blown into the work area 11. The air blown into the work area 11 is exhausted through holes in a punched panel on the floor surface 16 of the work area, cooled by the air cooling means 3, and then passes through the return shaft and returns to the ceiling supply chamber 14. The air that has returned to the ceiling supply chamber 14 is blown out again into the work area 11 by the conveying force of the fan filter unit (FFU) 2.

Although it is difficult to see in Figure 4, the volume of a large space (ballroom) clean room is large, and the fan filter unit 2, which is depicted abstractly in the figure, is not a single unit but a group, and the air flow area covered by three fan filter units 2 located on one side of one air cooling means 3 in the figure is imagined to be 25 to 30 m long for the three units. Also, multiple fan filter units 2 are imagined to be lined up on the depth side of Figure 4. In other words, the work area 11 is divided horizontally into multiple areas for each air cooling means 3, which is a dry coil, and air cooled by one air cooling means 3 circulates in the larger area.

The change in the indoor temperature in the work area 11 has been somewhat alleviated by the adoption of the ballroom system compared to previous systems. Even so, the tolerance for a set value of 23°C can be strict, for example at around ±1°C. This indoor temperature is designed to maintain the set value of 23°C anywhere in the room, even if production equipment or the like is in operation and there is heat generated in the room as a heat load. The equipment has the capacity to maintain 23°C after cooling the heat generated by the maximum load, that is, when the air leaves the exhaust port on floor 4 as return air. Therefore, the temperature of the air discharged from the ceiling will be much lower than 23°C if the indoor heat load is large, and will be close to 23°C if the load is small.

In this way, in conventional clean room air conditioning control systems, one air cooling means 3 is responsible for a large horizontally divided work area that is actually 25 to 30 m deep, and the flow rate of that air cooling means 3 is controlled by a single proportional control valve, so the flow rate of the heat transfer medium of the proportional control valve is controlled by a calculation signal based on the deviation between the measurement value of a single detection temperature sensor in the large work area and the temperature set value (for example, a set value of 23°C).

In conventional air conditioning control for clean rooms, a temperature sensor is installed near the side wall 22 of the return shaft, and the intake air temperature of the air cooling means (dry coil) 3 is adjusted by a temperature controller so that the measurement value of this temperature sensor becomes the target temperature, thereby keeping the air speed (fan rotation speed) of the fan filter unit 2 constant.
In general, temperature unevenness within the work volume area 11 is adjusted by an operator manually adjusting the wind speed setting (fan rotation speed) of the fan filter unit 2 .

Thus, in the temperature control method of conventional air conditioning control systems, in order to adjust temperature unevenness caused by differences in heat loads of production equipment installed in the work area, the operator had to manually adjust the air speed of the fan filter unit 2, which is divided into areas, and even if the layout of the production equipment was changed, the operator had to manually readjust the air speed of the fan filter unit 2.

Furthermore, in order to achieve energy savings, it was common for the operator to make fine adjustments to the wind speed of the fan filter unit 2 empirically while observing the current situation.

Air conditioning control systems are also known that use predictive methods such as regression control methods for temperature control. Examples of this type of prior art include Patent Document 1 and Patent Document 2.

Patent Literature 1 discloses a room temperature estimation device that makes it possible to estimate the room temperature of a room in a building from the measured outside air temperature without performing a computer simulation. This room temperature estimation device includes a memory unit, a prediction formula generation unit, a prediction trend acquisition unit, and a room temperature estimation unit, and the prediction formula generation unit uses the room temperature and outside air temperature data for each time at each of multiple times for multiple days in a predetermined extraction period stored in the memory unit to obtain multiple regression prediction formulas that each represent the relationship between the room temperature data and the outside air temperature data for each time as prediction formulas.
The room temperature estimation unit determines the outside temperature at a given date and time from the trend in outside temperature acquired by the predicted trend acquisition unit, and estimates the room temperature by fitting the outside temperature to a prediction formula for the time corresponding to that date and time.

Patent Document 2 discloses a water supply temperature control device that responds to changes in the characteristics of the heat source equipment and the circulation pump, and to changes in the external environment, and constantly determines the optimal water supply temperature over a long period of time. While the heat source equipment is in operation, this water supply temperature control device periodically collects and accumulates the amount of energy used by the heat source equipment 1 PW1, the amount of energy used by the cold/hot water pump PW2, the cold/hot water supply temperature TS from the heat source equipment, and the actual values of the outside air temperature tout as relevant parameters related to the current load situation.

Each time these related parameters are collected, the actual values of the collected related parameters are plotted in a multidimensional space, a response surface model is created using RSM-S technology, and the current optimal water supply temperature TSsp is determined from the response surface model that has been created.

Patent No. 6160945 Patent No. 5320128

However, Patent Document 1 is based on obtaining a regression equation that represents the room temperature using data on the room temperature and the outside air temperature at specific times on multiple days, and does not suggest temperature control in a clean room that is affected not only by disturbances in the outside air temperature but also by disturbances in heat generated by the production equipment inside the room. Patent Document 2 plots the past performance values of the cooling water supply temperature in a three-dimensional space with the total amount of energy used by the heat source equipment and the amount of energy used by the circulation pump on the first axis, the water supply temperature on the second axis, and the outside air temperature on the third axis, and cuts out a cross section of a response surface model created using multidimensional spline interpolation technology at the current outside air temperature. It determines the water supply temperature at which the total energy on this cross section is the minimum, and like the invention disclosed in Patent Document 1, does not suggest temperature control in a clean room that is affected not only by disturbances in the outside air temperature but also by disturbances in heat generated by the production equipment inside the room.

In view of the above, the object of the present invention is to provide an air conditioning control system that optimizes the temperature and saves energy in a clean room (including the work area) that is affected by heat disturbances from indoor production equipment, and that is automated without the need for operator intervention.

The inventors of the present application conducted an experiment using machine learning AI to make decisions while predicting results from past conditions, as a means of achieving so-called multi-objective optimization of energy-intensive cleanroom air conditioning that simultaneously achieves high cleanliness, temperature controllability, and energy conservation. The air conditioning control system of the present invention is the embodiment of this experiment.

The present invention has the following representative configuration. Note that, in order to facilitate understanding of the present invention, reference symbols used in the drawings of the embodiments described below are used here (Fig. 1), but the present invention is not limited to the configurations indicated by these reference symbols.

The present invention comprises: (1) a fan filter unit installed in a supply chamber provided in the ceiling of a work area inside a clean room, the work area being disposed on the side of which is a loading/unloading unit of a production device, the fan filter unit supplying conditioned air to the work area; and a return path provided from the side of the work area toward the end of the main body facing the loading/unloading unit of the production device and toward a maintenance passage, the return path causing the conditioned air supplied to the work area to pick up heat after passing between the production devices and become an updraft, and returning the heated conditioned air to the supply chamber.
An air conditioning control system including an air cooling means provided at a position where the conditioned air returning through the return path passes and cooling the air returning to the supply chamber,
an intake air temperature sensor for measuring the indoor temperature of the working area and controlling the air cooling means;
A plurality of return air temperature sensors which are wireless slave devices for measuring the temperature of the air passing through the return path and controlling the supply air volume of the fan filter and the air cooling means;
A system control unit having a programmable controller, a fan filter unit controller, a return air temperature sensor master unit that wirelessly receives measurement signals from the plurality of return air temperature sensors, and an AI PC;
a fan motor rotation speed controller for increasing or decreasing the amount of air supplied to the fan filter unit by a control signal from the system control unit;
a two-way valve for controlling the amount of cold water supplied to the air cooling means by a control signal from the system control unit;
The AI PC has machine learning software installed, and the AI PC outputs to the programmable controller at predetermined time intervals an optimal setting target value set, which is a set of optimal air temperature setting values passing through a plurality of return paths and an optimal indoor temperature setting value, based on the measurement value of the supply air temperature sensor, the measurement value of the return air temperature sensor, the current fan motor rotation speed operation amount, the current two-way valve opening amount, and the indoor temperature setting value;
The programmable controller resets the optimal setting target value set received from the AI PC as the set value for the air temperature passing through multiple return paths and the set value for the indoor temperature, receives the measurement value of the supply air temperature sensor and the measurement value of the return air temperature sensor, and PID controls the operation amount of the fan motor rotation speed controller and the two-way valve opening degree for increasing or decreasing the amount of air supplied to the fan filter unit.

(2) A fan filter unit is installed in a supply chamber provided in the ceiling of a work area inside the clean room, on the side of which the loading and unloading unit of the production equipment is located, and supplies conditioned air to the work area; and a return path is provided from the side of the work area toward the end of the main body facing the loading and unloading unit of the production equipment and further toward the maintenance passage, in which the conditioned air supplied to the work area becomes heated after passing between the production equipment and becomes an updraft, and the heated conditioned air is returned to the supply chamber.
An air conditioning control system including an air cooling means provided at a position where the conditioned air returning through the return path passes and cooling the air returning to the supply chamber,
an intake air temperature sensor for measuring the indoor temperature of the working area and controlling the air cooling means;
A plurality of return air temperature sensors which are wireless slave devices for measuring the temperature of the air passing through the return path and controlling the supply air volume of the fan filter and the air cooling means;
A system control unit having a programmable controller, a fan filter unit controller, a return air temperature sensor master unit that wirelessly receives measurement signals from the plurality of return air temperature sensors, and an AI PC;
a fan motor rotation speed controller for increasing or decreasing the amount of air supplied to the fan filter unit by a control signal from the system control unit;
a two-way valve for controlling the amount of cold water supplied to the air cooling means by a control signal from the system control unit;
The AI PC has machine learning software installed, and the AI PC outputs to the programmable controller at predetermined time intervals an optimal setting target value set, which is a set of an optimal setting value of the air temperature passing through a plurality of return paths, a gain related to a proportional band, and an offset related to an integral action, and an optimal setting value of the indoor temperature, a gain related to a proportional band, and an offset related to an integral action, based on the measurement value of the supply air temperature sensor, the measurement value of the return air temperature sensor, the current fan motor rotation speed operation amount, the current two-way valve opening amount, and the indoor temperature setting value,
The programmable controller resets the optimal setting target value set received from the AI PC as a set value for the air temperature passing through multiple return paths, a gain related to the proportional band, and an offset related to the integral action, and a set value for the indoor temperature, a gain related to the proportional band, and an offset related to the integral action, and then receives the measurement value of the supply air temperature sensor and the measurement value of the return air temperature sensor, and PID controls the operation amount of a fan motor rotation speed controller and a two-way valve opening degree for increasing or decreasing the amount of air supplied to the fan filter unit.

(3) The AI PC includes a regression prediction unit, an optimum setting unit, and a preferred solution selection unit,
The optimum setting unit is configured to have a plurality of objectives, such as maintaining a constant indoor temperature while improving energy saving performance, and receives the measurement value of the supply air temperature sensor, the measurement value of the return air temperature sensor, the current fan motor rotation speed operation amount, the current two-way valve opening amount, and the indoor temperature setting value, and presents an optimum setting target value set as a combination that achieves the plurality of objectives within a predetermined range;
After receiving the initial optimal setting target value set from the optimal setting unit, the regression prediction unit inputs the measurement values of the supply air temperature sensor and the multiple return air temperature sensors and the indoor temperature set value to calculate multiple candidate values for the optimal setting target value using a response surface provided in the regression prediction unit, predicts a control execution result from the multiple calculated candidate values for the optimal setting target value using a regression method, and passes these to the optimal setting unit again as predicted values of the control execution result, namely the measured predicted value of the supply air temperature sensor and the measured predicted value of the return air temperature sensor, the current fan motor rotation speed operation amount predicted value, the current two-way valve opening predicted value, the supply air temperature set value, and the indoor temperature set value,
The optimum setting unit presents a set of optimum setting target values based on the predicted values of the supply air temperature sensor and the predicted value of the return air temperature sensor, the current predicted value of the fan motor rotation speed operation amount, the current predicted value of the two-way valve opening, the supply air temperature set value, and the indoor temperature set value, as a combination that achieves the multiple objectives within a predetermined range, to the regression prediction unit again;
The air conditioning control system is characterized in that the air conditioning control system is as described in (1) or (2) above, in which the optimal solution selection unit selects one of the most suitable results from the multiple candidate values generated by evaluation using the optimal setting unit and the regression prediction unit based on a single-purpose selection rule, and the selected result is used as the optimal setting target value set.

(4) The AI PC further includes a database unit, a regression prediction creation unit, and a digital communication unit.
The database unit stores in advance virtual air conditioning control results obtained by virtually executing a large number of combinations of setting values created by a simulation prior to system operation, and the optimal setting unit and the regression prediction unit evaluate and generate multiple candidate values, and the optimal solution selection unit selects from the multiple candidate values based on a single-purpose selection rule, and stores the most optimal result from the previous actual operation, as well as the actual operation results.
The air conditioning control system is characterized in that the regression prediction creation unit creates a new regression prediction, which is a new response surface, using the virtual results and the actual operation results, and updates the response surface of the regression prediction unit using the created new regression prediction.

(5) The programmable controller is an air conditioning control system as described in (3), which includes a PID control section and performs PID calculations based on the optimal setting target value set obtained from the AI PC and the current measurement values of the supply air temperature sensor and the return air temperature sensor, and outputs the results to the fan filter unit controller and the two-way valve as the fan motor rotation speed operation amount of the fan filter unit and the two-way valve operation value.

(6) The clean room has a plurality of work areas arranged adjacent to each other in one direction in a plan view,
The supply chamber and the return path are disposed above the working area,
The air cooling means and the two-way valve are installed in common to the upper supply chambers of the plurality of work zones;
The air conditioning control system is characterized in that the supply air temperature sensor is installed in one of the work areas, and a plurality of return air temperature sensors are installed in each of the plurality of return paths of the work area.

(7) The air conditioning control system according to (1) or (2) is characterized in that the opening degree of the two-way valve is controlled by PID control based on the measurement value of the supply air temperature sensor so that the supply air temperature becomes the supply air temperature target value output by the AI PC.

The present invention is not limited to the above configuration or the configuration described in the embodiment described below, and it goes without saying that various modifications are possible within the scope of the technical concept of the present invention.

The air conditioning control system configured according to the present invention eliminates the need for the air conditioning manager (operator) to manually adjust the airflow settings of the fan filter units in each area to equalize the temperature variations caused by differences in heat load between areas.

AI installed in a PC constantly monitors the temperature conditions in the clean room (work area), making it possible to simultaneously achieve temperature control and energy savings.

By using AI to predict the results of control execution of numerous combinatorial optimization setting candidates using regression methods, the stability of the control results is improved. In addition, by periodically updating the regression prediction section with actual operation data, it becomes possible to deal with fluctuations in the indoor air conditioning load caused by the operating status of production equipment installed in the clean room.

FIG. 1 is a schematic diagram illustrating a basic configuration of an air conditioning control system according to the present invention. FIG. 1 is a functional block diagram illustrating a control system of an air conditioning control system according to the present invention. FIG. 1 is a plan view illustrating an example of the layout of work areas in a clean room building having multiple work areas, in which the air conditioning control system according to the present invention is implemented. Schematic diagram illustrating an example of the configuration of a conventional clean room.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an air conditioning control system according to the present invention will be described in detail below with reference to the drawings.
FIG. 1 is a schematic diagram explaining the basic configuration of an air conditioning control system according to the present invention, in which the loading and unloading section of production equipment 1 is arranged facing the interior enclosed by a ceiling 21, side walls 22, and slab floor 23 of a clean room building 20, and there is a work area 11 above the center and on the floor where products move back and forth. This work area 11 is an indoor area, and conditioned air circulates within the clean room, including a return air duct 13 described below.
A medium for cooling the circulating air-conditioning air, here cold water (shown as "cold water" in FIG. 1), is supplied from the outside through a cold water passage 40 into the tube of the air cooling means (dry coil) 3 via a two-way valve 4 that is electromagnetically or motor-driven, which will be described later. Reference numeral 30 denotes a system control unit.

Inside the clean room building 20, a supply chamber 14 is provided above the ceiling of the double ceiling 15 of the work area 11, which is a space targeted for a clean atmosphere and in which the production equipment 1 is placed. A fan filter unit 2 is installed in this supply chamber 14 and supplies cleaned conditioned air back to the work area 11, sealing the air outlet surface of the double ceiling 15.

The conditioned air passes between the multiple production devices 1 installed on top of the closed double floor 16 from the work area 11, and is heated while removing heat from the production devices, and then flows back to the supply chamber 14 through the return path 13. This return path 13 is formed between a partition wall 12 installed hanging down from the ceiling of the work area 11 and the side wall 13 of the clean room building 20. In FIG. 1, the return path 13 is provided on both sides of the work area 11. The partition wall 12 can be adjusted in how much it hangs down from the floor depending on the indoor conditions such as the size and position of the production devices 1 installed in the work area 11, and can also be installed hanging down to a position where it touches the upper end of the production devices 1.

The conditioned air that is circulated through the return path 13 rises in temperature and reaches the vertical end face of the supply chamber 14. An air cooling means (dry coil) 3 that cools the circulated air is installed on the vertical end face of the supply chamber 14, and the conditioned air is circulated so that it comes into contact with the outer surface of the tube and between the numerous fins fitted to the tube, and heat exchange occurs, cooling the conditioned air and adjusting its temperature.

A supply air temperature sensor 5 for measuring the indoor temperature is provided in a hollow position in the work area 11, and the measured supply air temperature is a value related to the opening degree of the two-way valve 4 of the air cooling means 3. In this embodiment, the supply air temperature sensor 5 is adapted to transmit the measured value to the system control unit 30 by wire. In addition, a return air temperature sensor 6 for measuring the temperature of the heated conditioned air passing through is provided in each of the multiple return paths 13 divided in the longitudinal direction of the elongated work area 11, and the measured return air temperature is a value related to the supply air volume of the fan filter units 2 grouped for each divided return path 13 and the control of the air cooling means. In this embodiment, the return air temperature sensor 6 is adapted to transmit the measured value to the system control unit 30 by wireless.

The reason why the return air temperature sensor 6 transmits its measured value wirelessly to the system control unit 30 is mainly because, as shown in the embodiment described below, in cases where the work area 11 is long and narrow and there are many divided return paths 13 (wide in area and far from the managed area), installation is easier and construction costs are lower than with wired installation. However, this does not deny the possibility of using wired transmission for the return air temperature sensor 6. Furthermore, this does not exclude the possibility of connecting the supply air temperature sensor 5 to the system control unit 30 wirelessly.

The system control unit 30 has a programmable controller 7, a fan filter unit controller 8, a return air temperature sensor master unit 9, and an AI PC 10. It also has a two-way valve 4 for adjusting the amount of cold water supplied to the air cooling means 3 with a control signal from the system control unit 30. Reference numeral 31 denotes a signal line for transferring measurement data and control data.

2 is a functional block diagram illustrating the control system of the air conditioning control system according to the present invention. The control procedure and machine learning software of this system are installed in the AI PC 10, and the rotation speed of the fan motor constituting the fan filter unit 2 is PID-controlled via the fan filter unit controller 8, with the optimal predicted value based on the measured values of the supply air temperature sensor 5 and the return air temperature sensor 6 and the indoor temperature setting value as the target value.
Alternatively, the AI PC 10 outputs an optimal setting target value set, which is a set of an optimal air temperature set value passing through the multiple return paths 13 and an optimal indoor temperature set value, based on the measurement value of the supply air temperature sensor 5, the measurement value of the return air temperature sensor 8, the current fan motor rotation speed operation amount, the current opening of the two-way valve 4, and the indoor temperature set value, to the programmable controller 7 described later at predetermined time intervals, and the programmable controller 7 resets the optimal setting target value set received from the AI PC 10 as the set value of the air temperature passing through the multiple return paths 13 and the set value of the indoor temperature, then receives the measurement value of the supply air temperature sensor 5 and the measurement value of the return air temperature sensor 8, and PID controls the operation amount of the fan motor rotation speed operation device and the two-way valve opening for increasing or decreasing the amount of air supply to the fan filter unit 2.

The AI PC (hereafter referred to as AI PC) 10 comprises a regression prediction unit 103, an optimum setting unit 102, and a preferred solution selection unit 101. The optimum setting unit 102 receives the current room temperature value from the return air temperature sensor 6, the current supply air temperature value and room temperature set value from the supply air temperature sensor 5, the current two-way valve 4 opening value, and the current FFU rotation speed, all of which are obtained from the digital communication unit 106, and proposes an optimum value from within the possible range of set values of the optimal air temperature set value passing through the multiple return paths 13 and the optimal room temperature set value, which have been input in advance. In this optimum setting unit 102, the air conditioning manager sets multiple objectives (for example, maintaining a constant room temperature while simultaneously increasing energy conservation, etc.), and outputs optimal setting values that satisfy the multiple objectives.

The set value (temperature set value) proposed by the optimum setting unit 102 is predicted by the regression prediction unit 103 when this setting is controlled and executed, and a predicted value is obtained. This predicted effective value is then evaluated again by the optimum setting unit 10.

To explain this in detail,
[1] The optimum setting unit 102 receives the current room temperature value, the current supply air temperature value, the room temperature set value, the current two-way valve 4 opening value, and the current FFU rotation speed value, and proposes a temperature set value that is considered to be good to the regression prediction unit 103. The regression prediction unit 103 then outputs the prediction result of executing the received set values.

[2] The optimum setting unit 102 receives the prediction result output from the regression prediction unit 103 and proposes a new optimum value to the regression prediction unit 103. The regression prediction unit 103 then outputs the prediction result obtained by executing the received new setting value.

In this manner, the optimum setting unit 102 proposes one optimum setting value to the regression prediction unit 103, and the regression prediction unit 103 outputs a predicted effective value obtained by executing the received setting value.
This predicted effective value is evaluated again by the optimum setting section 102, which then proposes one further optimum setting value to the regression prediction section 103, which then outputs a prediction result obtained by executing the received setting value.

In this way, the proposed settings and predicted values for the control execution results are repeated for hundreds of patterns, and a combination of the set values and predicted values is obtained for the number of repetitions (a set of hundreds of proposed settings and predicted values for the control execution results) ("A and B" in Figure 2).

The above several hundred combinations of set values and predicted values are sent to the preferred solution selection unit 101 (C), which applies single-purpose selection rules to select one set of set values. This set of set values is sent to the PLC 7 via the digital communication unit 106 and is periodically reflected in the air conditioning control of the clean room.

The preferred solution selection unit 101 selects one of the most preferred results from the multiple candidate values as an optimal predicted value, and provides the selected result as a target value to the programmable controller 7 via the digital communication unit 106.
After receiving the initial optimal setting target value set from the optimal setting unit 102, the regression prediction unit 103 inputs the measured values of the supply air temperature sensor 5 and the multiple return air temperature sensors 6 and the indoor temperature setting value to calculate multiple candidate values of the optimal setting target value using a response surface provided in the regression prediction unit 103, predicts the control execution result from the calculated multiple candidate values of the optimal setting target value using a regression method, and calculates the predicted values of the control execution result, i.e., the measured predicted value of the supply air temperature sensor 5 and the measured predicted value of the multiple return air temperature sensors 6, and the current multiple The optimum setting unit 102 again passes these values to the optimum setting unit 102 as the fan motor rotation speed operation variable predicted value, the current two-way valve opening predicted value, the supply air temperature set value, and the indoor temperature set value, and the optimum setting unit 102 again presents to the regression prediction unit 103 an optimum setting target value set based on the predicted values, which are the measurement predicted value of the supply air temperature sensor 5 and the measurement predicted value of the return air temperature sensor 6, the current fan motor rotation speed operation variable predicted value, the current two-way valve opening predicted value, the supply air temperature set value, and the indoor temperature set value, as a combination that achieves the multiple objectives within a predetermined range,
The optimal setting unit 102 and the regression prediction unit 103 evaluate and generate multiple candidate values, and the optimal solution selection unit 101 selects one of the multiple candidate values based on a single-purpose selection rule, and the selected result is set as the optimal setting target value set.

The AI PC 10 is equipped with a database 104. It also has a regression prediction creation unit 105 that periodically updates the regression prediction unit 103 from past data. The database unit 104 stores in advance virtual results and actual operation results obtained by virtually executing a large number of combinations of setting values created by a thermal fluid simulation prior to system operation. Setting values can be created in advance and predicted using this data.
Specifically, the database unit 104 pre-stores virtual air conditioning control results obtained by virtually executing a large number of combinations of setting values created in a simulation prior to system operation, and multiple candidate values generated by evaluation by the optimum setting unit 102 and the regression prediction unit 103 are added and stored as needed along with the actual operation result up to the previous time, which is the most suitable result selected from the multiple candidate values by the preferred solution selection unit 101 based on a single-purpose selection rule, and the regression prediction creation unit 105 creates a new regression prediction, which is a new response surface, using the virtual results and the actual operation results, and updates the response surface of the regression prediction unit 103 using the new regression prediction created.

This configuration makes it possible to obtain a regression-based predicted value of the result of executing setting values that are not adopted in normal operation and therefore cannot be obtained in actual operation.
In addition, by adding actual operation results, the current situation is monitored from both the simulation results and the actual operation results, and the appropriate regression prediction unit 103 is updated periodically (for example, once a day). In this way, the regression prediction unit is changed according to the actual operation situation. The setting value output by the AIPC is not only the temperature setting value, but also the gain and offset can be output.

The programmable controller 7 is equipped with a PID control section 71, and executes PID calculations based on the target value obtained from the AI PC 10 and the current measured values of the supply air temperature sensor 5 and the return air temperature sensor 6, and outputs the result to the fan filter unit controller 8 as the operation value of the fan filter unit 2. The fan filter unit controller 8 issues a control signal to a group of multiple fan filter units 2 installed in the unit work area 11.

The programmable controller 7 receives temperature data from the return air temperature sensor parent unit 9 and temperature data from the supply air temperature sensor 5, and also receives the target value from the AI PC 10. The target value obtained from the AI PC 10 is not only the temperature setting value, but also the setting value including the gain and offset, and can be instructed to the PID control unit.

The programmable controller 7 also has an internal PID control section, which obtains the target value obtained from the AI PC 10 and the current values from each temperature sensor (supply air temperature sensor 5, return air temperature sensor 6) and outputs an operation value through PID calculation. If a fan filter unit controller 8 is provided separately, the rotation speed of each fan filter unit 2 group is output to the fan filter unit controller 8 for indirect control.

An operation value is output directly to the dry coil two-way valve 4 without going through a separate controller.
In addition, the programmable controller 7 periodically transmits to the AI PC 10 the current room temperature etc. value (return air temperature sensor 6) from the return air temperature sensor parent unit 9, the current temperature value from the supply air temperature sensor 5, the indoor temperature setting value set by the air conditioning manager, the rotation speed instruction value of the fan filter unit 2 group to be transmitted, and the operation instruction value for the dry coil two-way valve 4.

The analog data of the PID control unit 71 is the input of the measured value of the supply air temperature sensor 5 transmitted by wire and the opening control signal of the two-way valve 4, and is carried out via the analog communication unit 72. The target value output from the digital communication unit 106 of the AI PC 10 is transferred to the PID control unit 71 by the digital communication unit 73 of the programmable controller 7. Communication between the PID control unit 71, the fan filter unit controller roller 8, and the parent unit 9 of the multiple return air temperature sensors 6 is also carried out via the digital communication unit 73.

In this way, the programmable controller 7 receives the temperature measurement value from the return air temperature sensor parent unit 9 and the temperature measurement value from the supply air temperature sensor 5, and receives the target value from the AI PC 10. The target value obtained from the AI PC 10 can include not only the temperature setting value, but also the gain and offset, and can be instructed to the PID control unit 71.

The PID control unit 71 in the programmable controller 7 outputs an operation value through PID calculation using the target value obtained from the AI PC 10 and the current measurement values obtained from the temperature sensors 5 and 6. The operation value is output directly from the programmable controller 7 to the two-way valve 4 that supplies cold water to the dry coil 3.

The programmable controller 7 periodically transmits the current temperature measurement value from the return air temperature sensor parent unit 9, the current temperature measurement value from the supply air temperature sensor 5, the rotation speed instruction value of the group of fan filter units 2 to be transmitted, and the operation instruction value for the two-way valve 4 to the AI PC 10.

Figure 3 is a plan view illustrating an example of a work area arrangement in which the air conditioning control system according to the present invention is implemented in a large-space clean room building with multiple work areas. In the clean room building 20, multiple work areas 11 are arranged adjacent to each other in two directions in a plan view. Each work area 11 has the supply chamber 14 and return path 13, and the air cooling means 3 and two-way valve 4 are installed in common to the multiple work areas 11.

The supply air temperature sensor 5 is installed in one of the multiple work areas 11, and a return air sensor 6 is installed in each of the return paths of the multiple work areas 11.

The opening of the two-way valve 4 is controlled by PID control based on the measurement value of the supply air temperature sensor 5 and the target value.

As already explained, the supply air temperature sensor 5 is a sensor that transmits temperature measurements to the system control unit 30 via a wire, and the return air temperature sensor 6 is a sensor that transmits temperature measurements to the system control unit 30 wirelessly. The reason for using a wireless sensor for the return air temperature sensor 6 is that it does not require the installation of communication lines and can easily accommodate changes in the installation location.

The type and location of the production equipment installed in the work area are not limited to the uniform one shown in FIG. 3. The return air temperature sensor 6 is placed according to the type, number, and layout of the production equipment.

The present invention is not limited to the above-mentioned industrial clean rooms, but can also be applied to warehouses for agricultural products, seafood, and beverages that require constant temperature control.

1... Production equipment 2... Fan filter unit (FFU)
3...Air cooling means (dry coil)
4: Two-way valve 5: Supply air temperature sensor 6: Return air temperature sensor 7: Programmable controller (PLC)
8...Fan filter unit controller (FFU controller)
9: Return air temperature sensor parent unit 10: AI PC (PC with machine learning software installed)
11...Work area (area for maintaining a clean atmosphere)
12: bulkhead 13: return path (maintenance area)
Reference Signs List 14: Supply chamber provided above the ceiling 15: Double ceiling 16: Floor of work area 20: Clean room building 21: Ceiling of clean room building 22: Side wall of clean room building 23: Floor of clean room building 30: System control unit 31: Signal path 40: Cold water path

Claims (7)

a fan filter unit installed in a supply chamber provided in the ceiling of a work area in the clean room, the work area being disposed on the side thereof with a loading/unloading unit of a production device, the fan filter unit supplying conditioned air to the work area; a return path provided from the side of the work area toward the end of the main body facing the loading/unloading unit of the production device and toward the maintenance passage, the return path causing the conditioned air supplied to the work area to pick up heat after passing between the production devices and become an updraft, and returning the heated conditioned air to the supply chamber;
An air conditioning control system including an air cooling means provided at a position where the conditioned air returning through the return path passes and cooling the air returning to the supply chamber,
an intake air temperature sensor for measuring the indoor temperature of the working area and controlling the air cooling means;
A plurality of return air temperature sensors which are wireless slave devices for measuring the temperature of the air passing through the return path and controlling the supply air volume of the fan filter and the air cooling means;
A system control unit having a programmable controller, a fan filter unit controller, a return air temperature sensor master unit that wirelessly receives measurement signals from the plurality of return air temperature sensors, and an AI PC;
a fan motor rotation speed controller for increasing or decreasing the amount of air supplied to the fan filter unit by a control signal from the system control unit;
a two-way valve for controlling the amount of cold water supplied to the air cooling means by a control signal from the system control unit;
The AI PC has machine learning software installed, and the AI PC outputs to the programmable controller at predetermined time intervals an optimal setting target value set, which is a set of optimal air temperature setting values passing through a plurality of return paths and an optimal indoor temperature setting value, based on the measurement value of the supply air temperature sensor, the measurement value of the return air temperature sensor, the current fan motor rotation speed operation amount, the current two-way valve opening amount, and the indoor temperature setting value;
The programmable controller resets the optimal setting target value set received from the AI PC as the set value for the air temperature passing through multiple return paths and the set value for the indoor temperature, receives the measurement values of the supply air temperature sensor and the return air temperature sensor, and PID controls the operation amount of the fan motor rotation speed controller and the two-way valve opening degree for increasing or decreasing the amount of air supplied to the fan filter unit. This air conditioning control system is characterized by the above. a fan filter unit installed in a supply chamber provided in the ceiling of a work area in the clean room, the work area being disposed on the side thereof with a loading/unloading unit of a production device, the fan filter unit supplying conditioned air to the work area; a return path provided from the side of the work area toward the end of the main body facing the loading/unloading unit of the production device and toward the maintenance passage, the return path causing the conditioned air supplied to the work area to pick up heat after passing between the production devices and become an updraft, and returning the heated conditioned air to the supply chamber;
An air conditioning control system including an air cooling means provided at a position where the conditioned air returning through the return path passes and cooling the air returning to the supply chamber,
an intake air temperature sensor for measuring the indoor temperature of the working area and controlling the air cooling means;
A plurality of return air temperature sensors which are wireless slave devices for measuring the temperature of the air passing through the return path and controlling the supply air volume of the fan filter and the air cooling means;
A system control unit having a programmable controller, a fan filter unit controller, a return air temperature sensor master unit that wirelessly receives measurement signals from the plurality of return air temperature sensors, and an AI PC;
a fan motor rotation speed controller for increasing or decreasing the amount of air supplied to the fan filter unit by a control signal from the system control unit;
a two-way valve for controlling the amount of cold water supplied to the air cooling means by a control signal from the system control unit;
The AI PC has machine learning software installed, and the AI PC outputs to the programmable controller at predetermined time intervals an optimal setting target value set, which is a set of an optimal setting value of the air temperature passing through a plurality of return paths, a gain related to a proportional band, and an offset related to an integral action, and an optimal setting value of the indoor temperature, a gain related to a proportional band, and an offset related to an integral action, based on the measurement value of the supply air temperature sensor, the measurement value of the return air temperature sensor, the current fan motor rotation speed operation amount, the current two-way valve opening amount, and the indoor temperature setting value,
The programmable controller resets the optimal setting target value set received from the AI PC as the set value of the air temperature passing through multiple return paths, the gain related to the proportional band, and the offset related to the integral action, and the set value of the indoor temperature, the gain related to the proportional band, and the offset related to the integral action, and then receives the measurement values of the supply air temperature sensor and the return air temperature sensor, and PID controls the operation amount of the fan motor rotation speed controller and the two-way valve opening degree for increasing or decreasing the amount of air supplied to the fan filter unit. The AI PC includes a regression prediction unit, an optimum setting unit, and a suitable solution selection unit,
The optimum setting unit is configured to have a plurality of objectives, such as maintaining a constant indoor temperature while improving energy saving performance, and receives the measurement value of the supply air temperature sensor, the measurement value of the return air temperature sensor, the current fan motor rotation speed operation amount, the current two-way valve opening amount, and the indoor temperature setting value, and presents an optimum setting target value set as a combination that achieves the plurality of objectives within a predetermined range;
After receiving the initial optimal setting target value set from the optimal setting unit, the regression prediction unit inputs the measurement values of the supply air temperature sensor and the multiple return air temperature sensors and the indoor temperature set value to calculate multiple candidate values for the optimal setting target value using a response surface provided in the regression prediction unit, predicts a control execution result from the multiple calculated candidate values for the optimal setting target value using a regression method, and passes these to the optimal setting unit again as predicted values of the control execution result, namely the measured predicted value of the supply air temperature sensor and the measured predicted value of the return air temperature sensor, the current fan motor rotation speed operation amount predicted value, the current two-way valve opening predicted value, the supply air temperature set value, and the indoor temperature set value,
The optimum setting unit presents a set of optimum setting target values based on the predicted values of the supply air temperature sensor and the predicted value of the return air temperature sensor, the current predicted value of the fan motor rotation speed operation amount, the current predicted value of the two-way valve opening, the supply air temperature set value, and the indoor temperature set value, as a combination that achieves the multiple objectives within a predetermined range, to the regression prediction unit again;
The air conditioning control system according to claim 1 or 2, characterized in that the optimal solution selection unit selects one of the most suitable results from multiple candidate values generated by evaluation by the optimal setting unit and the regression prediction unit based on a single-purpose selection rule, and the selected result is set as the optimal setting target value set. The AI PC further includes a database unit, a regression prediction creation unit, and a digital communication unit,
The database unit stores in advance virtual air conditioning control results obtained by virtually executing a large number of combinations of setting values created by a simulation prior to system operation, and the optimal setting unit and the regression prediction unit evaluate and generate multiple candidate values, and the optimal solution selection unit selects from the multiple candidate values based on a single-purpose selection rule, and stores the most optimal result from the previous actual operation, as well as the actual operation results.
4. The air conditioning control system according to claim 3, wherein the regression prediction creation unit creates a new regression prediction, which is a new response surface, using the virtual results and the actual operation results, and updates the response surface of the regression prediction unit using the created new regression prediction. The air conditioning control system according to claim 1 or 2, characterized in that the programmable controller is equipped with a PID control section, and performs PID calculations based on the optimal setting target value set obtained from the AI PC and the current measurement values of the supply air temperature sensor and the return air temperature sensor, and outputs the results to the fan filter unit controller and the two-way valve as the fan motor rotation speed operation amount and the two-way valve operation value of the fan filter unit. The clean room has a plurality of working areas arranged adjacent to each other in one direction in a plan view,
The supply chamber and the return path are disposed above the working area,
The air cooling means and the two-way valve are installed in common to the upper supply chambers of the plurality of work zones;
The air conditioning control system according to any one of claims 1 to 5, characterized in that the supply air temperature sensor is installed in one of the work areas, and a plurality of return air temperature sensors are installed in each of the plurality of return paths of the work area. The air conditioning control system according to claim 1 or 2, characterized in that the opening degree of the two-way valve is controlled by PID control based on the measurement value of the supply air temperature sensor so that the supply air temperature becomes the supply air temperature target value output by the AI PC.

Images