An energy-saving control method and device for a fresh air system

Abstract

The application discloses an energy-saving control method and device for a fresh air system. The energy-saving control method comprises the following steps: obtaining current environmental detection data of a target area, and inputting the historical environmental detection data into a preset LSTM model to predict a personnel density change rate in a short future, so as to output a short-term control instruction to control the short-term output air volume of the fresh air system; inputting the personnel density change rate in the short future and the current environmental detection data into a layered DRL controller to output a long-term control instruction to control the long-term output air volume of the fresh air system; obtaining an execution judgment result of the fresh air system based on a preset judgment logic, and obtaining an air volume control result based on the execution judgment result, so as to control the fan rotating speed, the valve opening degree and the bypass valve opening degree of the fresh air system according to the air volume control result. The energy-saving control method can dynamically balance the contradiction between the fresh air volume and the energy consumption, thereby reducing the energy consumption of the fresh air system and improving the overall operation efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to an energy-saving control method for a fresh air system, and also relates to a corresponding energy-saving control device, and belongs to the technical field of air conditioning. BACKGROUND

[0002] The fresh air system plays a key role in regulating indoor air quality in modern buildings, but its energy consumption usually accounts for 20% to 40% of the total building energy consumption, so it has become the focus of green building and energy-saving reconstruction. With the continuous improvement of energy-saving standards, how to effectively reduce the energy consumption of the fresh air system has become a hot issue in current research and engineering practice. The traditional control strategy mainly relies on fixed air volume or simple variable air volume regulation, and it is difficult to realize energy consumption optimization while ensuring air quality under the condition of drastic change of personnel density or extreme climate.

[0003] At present, the common fresh air control methods include four typical ways: the first is a fixed time start-stop control, that is, running according to a preset time table, which is simple to operate but cannot respond to sudden personnel changes; the second is a CO2 concentration feedback control, which adjusts the air volume through a PID algorithm, although it can respond to air quality changes, but it has the problems of response lag and easy to cause system oscillation; the third is a double air volume switching control, which only sets high and low air volume, and the energy-saving effect is limited, and it is difficult to adapt to complex scenes; the fourth is a linkage control based on PM2.5 / VOC, although the pollutant index is introduced, but the influence of heat load on system operation is ignored.

[0004] The above methods generally have three defects: first, the control logic relies on static threshold setting, and lacks dynamic adaptation to environmental changes; second, the multi-objective conflict between energy saving, air quality and human comfort cannot be effectively coordinated; third, the future state lacks the ability of forward-looking prediction, resulting in slow response of the system when the personnel density changes suddenly, and the air volume regulation is easy to oscillate, and the real-time collaborative optimization of building personnel density and external environmental parameters cannot be realized. SUMMARY

[0005] The primary technical problem to be solved by the present application is to provide an energy-saving control method for a fresh air system.

[0006] Another technical problem to be solved by the present application is to provide an energy-saving control device for a fresh air system.

[0007] To achieve the above technical purposes, the present application adopts the following technical solutions:

[0008] According to a first aspect of an embodiment of the present application, an energy-saving control method for a fresh air system is provided, comprising the following steps:

[0009] obtaining current environmental detection data of a target area;

[0010] combining the current environment detection data with historical environment detection data of the target area, inputting into a preset LSTM model to predict a personnel density change rate in a short future; wherein the LSTM model is trained based on the historical environment detection data of the target area;

[0011] outputting a short-term control instruction based on the personnel density change rate in the short future; wherein the short-term control instruction is used to directly control a short-term output air volume of the fresh air system;

[0012] inputting the personnel density change rate in the short future and the current environment detection data into a hierarchical DRL controller to output a long-term control instruction; wherein the long-term control instruction is used to adjust energy consumption weight, CO2 over-standard weight and comfort weight of the fresh air system to control a long-term output air volume of the fresh air system through weighted average;

[0013] obtaining an execution judgment result of the fresh air system based on preset judgment logic; wherein the execution judgment result is one of the short-term control instruction or the weighted control instruction of the short-term and the long-term;

[0014] obtaining an air volume control result of the fresh air system based on the execution judgment result, to control fan speed, valve opening degree and bypass valve opening degree of the fresh air system based on the air volume control result.

[0015] Preferably, the obtaining of the current environment detection data of the target area specifically includes:

[0016] obtaining CO2 concentration of the target area through a double CO2 sensor array deployed on a ceiling of the target area;

[0017] detecting current personnel density of the target area through an infrared detection device and a Wi-Fi detection device;

[0018] obtaining enthalpy difference value of the target area through an enthalpy difference calculation module deployed in a corridor of the target area;

[0019] obtaining outdoor temperature and humidity and PM2.5 detection result of the target area through an external environment sensor;

[0020] The CO2 concentration, the current personnel density, the enthalpy difference value, the outdoor temperature and humidity and the PM2.5 detection result of the target area jointly constitute the current environment detection data of the target area.

[0021] Preferably, the outputting of the short-term control instruction based on the personnel density change rate in the short future specifically includes:

[0022] If the personnel density change rate is greater than the first threshold value, the short-term control instruction is to quickly increase the air volume and close the bypass valve;

[0023] If the personnel density change rate is not greater than the first threshold value and is greater than the second threshold value, the short-term control instruction is to increase the air volume in a stepwise manner by increasing one air volume gradient every preset interval length;

[0024] If the personnel density change rate is not greater than the second threshold value and is greater than the third threshold value, the short-term control instruction is to maintain the current air volume and fine-tune the air volume by a preset amplitude;

[0025] If the personnel density change rate is not greater than the third threshold value, the short-term control instruction is to reduce the air volume and start the night cleaning mode;

[0026] Among them, the first threshold value > the second threshold value > the third threshold value, and the first threshold value and the second threshold value are both positive percentages, and the third threshold value is a negative percentage.

[0027] Preferably, the long-term control instruction is obtained by the following steps:

[0028] Based on the current environmental detection data, the outdoor temperature Tout is obtained; and the energy consumption weight α is calculated based on the temperature set value Tset and the temperature sensitive coefficient k, ;

[0029] Based on the personnel density change rate and the CO2 concentration of the target area, the CO2 over-standard weight β is calculated, ; wherein σ represents the personnel density sensitive coefficient; ΔOcc represents the personnel density change rate predicted by the LSTM model, and the value is 0-1; CO2 represents the CO2 concentration of the target area; λ represents the CO2 compensation coefficient; 800 represents the target threshold value of the CO2 concentration;

[0030] Based on the energy consumption weight α and the CO2 over-standard weight β, the comfort weight γ is calculated, γ = 1- α- β, and γ ≥ γ min ; wherein γ min represents the lower limit of the comfort weight of the target area; if γ ≥ γ min , the current weight is maintained; if γ < γ min , γ = γ min , and the energy consumption weight α and the CO2 over-standard weight β are compressed in proportion;

[0031] Based on the energy consumption weight α, the CO2 over-standard weight β and the comfort weight γ, the long-term output air volume of the fresh air system is output by weighted calculation.

[0032] Preferably, the long-term output air volume is weighted calculated by the following steps:

[0033] The energy consumption suggestion air volume is calculated based on the minimum energy consumption curve;

[0034] The air quality suggestion air volume is obtained based on the CO2 change rate;

[0035] According to the comfort level, the comfort level suggestion air volume required for maintaining the human thermal reaction PMV ∈ [-0.5, +0.5] is obtained;

[0036] The energy consumption suggestion air volume, the air quality suggestion air volume, and the comfort level suggestion air volume are normalized based on the maximum air volume of the fresh air system; wherein, energy = energy consumption suggestion air volume / system maximum air volume 100%, u air = air quality suggestion air volume / system maximum air volume 100%, u comfort = comfort level suggestion air volume / system maximum air volume 100%;

[0037] The final weighted calculation result u final of the fresh air system is obtained based on the normalization calculation result; ;

[0038] The long-term output air volume Q final of the fresh air system is calculated based on the final weighted calculation result u final ; wherein Q final = u final system maximum air volume.

[0039] More preferably, the preset judgment logic specifically includes:

[0040] The output air volume difference is calculated based on the short-term output air volume and the long-term output air volume;

[0041] The output air volume difference is compared with the air volume control maximum allowed deviation preset for the fresh air system;

[0042] If the output air volume difference is not greater than the air volume control maximum allowed deviation, the short-term output air volume and the long-term output air volume are weighted and averaged according to a preset proportion, and a weighted control instruction is executed based on the weighted average result;

[0043] If the output air volume difference is greater than the air volume control maximum allowed deviation, an arbitration mechanism is triggered, an arbitration result is output based on the arbitration mechanism, and a corresponding control instruction is executed based on the arbitration result; wherein the arbitration result is directly executing the short-term control instruction or executing a weighted control instruction synthesized based on the short-term control instruction and the long-term control instruction after adjusting the weight.

[0044] The arbitration mechanism comprises, in particular:

[0045] Conflict detection is performed based on the short-term control instruction and the long-term control instruction;

[0046] If there is a conflict, the short-term control instruction is executed preferentially;

[0047] If there is no conflict, it is determined whether CO2 exceeds the standard; if CO2 exceeds the standard, the weight of the short-term control instruction is increased by 30%, and if CO2 does not exceed the standard, the weight of the long-term control instruction is increased by 20% as the weighted control instruction after weight adjustment.

[0048] The energy-saving control method further comprises, in particular:

[0049] The Mahalanobis distance is used to detect the abnormal value of the sensor for the CO2 sensor array and the external environment sensor:

[0050] ;

[0051] wherein D M represents the Mahalanobis distance value, x represents the observation value vector of the current sensor, μ represents the mean value vector of the sensor data, represents the deviation of the current data from the normal state, represents the inverse matrix of the covariance matrix;

[0052] Based on the preset threshold and the abnormal value of each sensor, it is determined whether each sensor is failed;

[0053] If a single external environment sensor is failed, a redundant sensor is switched; if the sensor array is failed, a GAN data generator is enabled to replace the sensor array.

[0054] According to a second aspect of the embodiment of the present application, an energy-saving control device for a fresh air system is provided, comprising:

[0055] A data acquisition layer is configured to acquire current environmental detection data of a target area;

[0056] An edge computing layer is internally provided with a preset LSTM model and a hierarchical DRL controller, and is connected with the data acquisition layer; the edge computing layer is configured to predict the personnel density change rate in the future short term according to the LSTM model, so as to output a short-term control instruction; the edge computing layer is further configured to output a long-term control instruction according to the hierarchical DRL controller;

[0057] An execution control layer is connected with the edge computing layer and has preset judgment logic, so as to obtain an execution judgment result of the fresh air system based on the preset judgment logic, and obtain a wind volume control result of the fresh air system based on the execution judgment result, so as to control the fan speed, the valve opening degree and the bypass valve opening degree of the fresh air system based on the wind volume control result.

[0058] According to a third aspect of the embodiments of the present application, another energy-saving control device for a fresh air system is provided, which comprises a processor and a memory, and the processor reads a computer program in the memory, so as to implement the above-mentioned energy-saving control method.

[0059] Compared with the prior art, the present application has the following technical effects:

[0060] (1) By predicting the personnel density change rate in the future short term, a short-term control instruction is outputted, and combined with a long-term control instruction for joint control, so as to dynamically balance the contradiction between fresh air volume and energy consumption, thereby reducing the energy consumption of the fresh air system and improving the overall operation efficiency.

[0061] (2) By balancing the conflict between CO2 concentration and heat recovery efficiency, the duration of CO2 exceeding the standard can be significantly reduced, so that the peak value of CO2 can be controlled below 800ppm.

[0062] (3) By detecting the abnormality of the sensor, corresponding remedial measures can be taken according to the abnormal result, so that the fault tolerance of the energy-saving control method is improved, and the stability of the operation of the fresh air system is improved. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 A whole flow chart of an energy-saving control method for a fresh air system provided in the first embodiment of the present application;

[0064] Figure 2 An architecture diagram of an LSTM model in the first embodiment of the present application;

[0065] Figure 3 A structure diagram of a hierarchical DRL controller in the first embodiment of the present application;

[0066] Figure 4 A flow chart of calculating the comfort weight in the first embodiment of the present application;

[0067] Figure 5 A flow chart of the preset judgment logic in the first embodiment of the present application;

[0068] Figure 6 A flow chart of the arbitration mechanism in the first embodiment of the present application;

[0069] Figure 7A structural diagram of an energy-saving control device for a fresh air system provided by a second embodiment of the present application is shown in the figure.

[0070] Figure 8 A structural diagram of an energy-saving control device for a fresh air system provided by a third embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0071] The technical content of the present application will be described in detail below in combination with the drawings and specific embodiments.

[0072] The embodiment of the present application applies multi-modal perception of external environment and hierarchical reinforcement learning to energy-saving control of fresh air system, aiming to solve the contradiction between fresh air volume and energy consumption that cannot be dynamically balanced by traditional control strategy, especially in the case of conflict between CO2 concentration control and heat recovery efficiency. Moreover, it can respond to changes in personnel density in the building and changes in external environment in real time, ensuring the overall ventilation effect of the fresh air system. In addition, it can also avoid the problem of energy waste caused by the surge of fresh air load under extreme weather conditions.

[0073] First embodiment

[0074] As shown in the figure, the energy-saving control method for a fresh air system provided by the first embodiment of the present application specifically includes the following steps: Figure 1

[0075] S10: Obtain the current environmental detection data of the target area.

[0076] Specifically, it includes steps S11-S14:

[0077] S11: Obtain the CO2 concentration of the target area.

[0078] Specifically, the target area in this embodiment can be an office (for example: office building), or a classroom or a venue, etc. The CO2 concentration of each sub-area is collected by a double CO2 sensor array installed on the ceiling of each sub-area. Moreover, the voting mechanism is used to exclude outliers, and finally the CO2 concentration of the target area is calculated by weighted average.

[0079] S12: Obtain the current personnel density of the target area.

[0080] First, the number of current Wi-Fi connections is detected by a Wi-Fi detection device to roughly estimate the real-time number of people; then, the infrared detection device detects human thermal signals using infrared thermal imaging principle to make up for the Wi-Fi blind area (for example: no mobile phone users), so as to adjust the estimated real-time number of people, and finally obtain the current personnel density of the target area based on the adjusted real-time number of people.

[0081] ​In addition, in this step, the number of people attending the scheduled meeting can also be obtained through the access control / conference system data, thereby assisting in predicting the personnel density change in the next 15-30 minutes.

[0082] S13: Obtain the enthalpy difference value of the target area.

[0083] Specifically, the enthalpy difference calculation module is deployed in the corridor of the target area, so as to obtain the enthalpy difference value of the target area through enthalpy difference calculation. The enthalpy difference calculation method is known in the art and will not be described in detail here.

[0084] S14: Obtain the outdoor temperature and humidity and PM2.5 detection results of the target area through external environment sensors.

[0085] It can be understood that the CO2 concentration, current personnel density, enthalpy difference value, outdoor temperature and humidity, and PM2.5 detection results of the target area together constitute the current environmental detection data of the target area.

[0086] S20: Predict the personnel density change rate in the near future.

[0087] After the current environmental detection data of the target area is collected through the above step S10, the current environmental detection data needs to be combined with the historical environmental detection data (in this embodiment, the historical detection data within 6 hours) of the target area, and input into the preset LSTM model together, to predict the personnel density change rate in the near future.

[0088] In this embodiment, the LSTM model is trained based on the historical environmental detection data of the target area. As shown in Figure 2 the architecture of the LSTM model includes:

[0089] ① Input layer: used for inputting 12-dimensional features.

[0090] Specifically, 12-dimensional features are obtained by preprocessing the historical environmental detection data of the target area, including outdoor dry-bulb temperature, indoor average temperature, indoor relative humidity, current CO2 concentration, current personnel density, CO2 change rate, PM2.5 detection results, meeting calendar flag, week feature, and time period feature, etc.

[0091] ② Hidden layer: 64 LSTM units, dropout = 0.2.

[0092] ③ Attention layer: Bahdanau attention mechanism.

[0093] ④ Output layer: predicted value of the personnel density change rate in the next 15-30 minutes.

[0094] It can be understood that after the model training is performed with the historical environment detection data of the target area as a training set, a preliminary model is formed; and a smooth L1 loss is used as a loss function, and an Adam optimizer is used for model optimization, and finally a pre-trained LSTM model is formed.

[0095] Therefore, when the current environment detection data of the target area is collected based on the above step S10, the pre-trained LSTM model is input with the historical data within 6 hours, so as to output the personnel density change rate in the next 15-30 minutes.

[0096] S30: output short-term control instructions.

[0097] Specifically, after the personnel density change rate ΔOcc in the next 15-30 minutes of the target area is obtained based on the above step S20, the personnel density change rate ΔOcc is compared with a plurality of preset threshold values, so as to output different short-term control instructions according to the comparison result. Specifically as follows:

[0098] (1) If ΔOcc>15% (i.e. the first threshold value), the air volume is quickly increased.

[0099] (2) If 5% (i.e. the second threshold value) < ΔOcc ≤ 15%, the air volume is increased in steps.

[0100] (3) If -10% (i.e. the third threshold value) < ΔOcc ≤ 5%, the air volume is maintained and fine-tuned.

[0101] (4) If ΔOcc ≤ -10%, the air volume is reduced and the night cleaning mode is started.

[0102] For example, in a specific embodiment, for the different types of short-term control instructions described above, the following execution strategies can be used: (1) quick increase of air volume: control the opening of the air valve to increase by 25%, the fan speed to increase by 20%, and the bypass valve to be closed; (2) stepwise increase of air volume: control the fan speed to increase by 10%, and adjust the air volume gradient by 5% every 30 seconds; (3) fine-tune the air volume: fine-tune the air volume with a range of ±2%; (4) reduce the air volume: reduce to 30% of the maximum air volume.

[0103] The night cleaning mode is an energy-saving ventilation strategy that runs at night or when the building is unoccupied, and the main purpose is to use the cool air at night to pre-cool the building structure and remove indoor pollutants.

[0104]

[0105] The initial control device sends instructions to the EC fan to reduce the speed to 30% of the maximum air volume. The control device adjusts the opening of the air valve to increase it to 90° (to increase the air flow passage). The electrostatic dust removal function of the filter screen is started, and the purification of particulate matter in the air is started.

[0106] Periodic self-cleaning phase (performed every 30 minutes). The control device briefly increases the speed of the EC fan to 80% of the maximum air volume and maintains it for 2 minutes. When the EC fan is running at high speed, it generates a high-pressure air flow that performs a backflush cleaning of the filter screen, blowing off the dust attached to the filter screen. After backflush cleaning, the filter screen detects the current pressure difference (reflecting the degree of blockage) and feeds back the data to the control device to monitor the status of the filter screen. This cycle is repeated every 30 minutes to ensure that the filter screen maintains high filtering capacity while avoiding the problem of dust accumulation caused by long-term low-speed operation.

[0107] It can be understood that in the embodiment, the short-term control instruction output based on the personnel density change rate in the short term is used to directly control the short-term output air volume of the fresh air system.

[0108] S40: Output long-term control instruction.

[0109] Specifically, based on the personnel density change rate in step S20, the current environmental detection data in step S10 is input into the hierarchical DRL controller to output a long-term control instruction. In the embodiment, as shown in the formula (1), the hierarchical DRL controller includes a state space, an action space, and a reward function. The state space is used to obtain the personnel density change rate and the current environmental detection data; the action space is used to allocate the energy consumption weight, the CO2 over-standard weight, and the comfort weight of the fresh air system; and the reward function is used to calculate the long-term output air volume of the fresh air system. Figure 3

[0110] As shown in the formula (2), the step S40 specifically includes the following steps: Figure 4

[0111] S41: Calculate the energy consumption weight α.

[0112] Specifically, based on the current environmental detection data, the outdoor temperature T out is obtained. And according to the temperature set value T set and the temperature sensitive coefficient k, the energy consumption weight α is calculated,

[0113] .

[0114] S42: Calculate the CO2 over-standard weight β.

[0115] ​​Specifically, based on the personnel density change rate ΔOcc and the CO2 concentration of the target area, the CO2 exceeding weight β is calculated together,

[0116] ;

[0117] Wherein, σ represents the personnel density sensitive coefficient, different coefficients are selected according to different target areas, for example: office scene: σ = 0.3 (personnel density influence is moderate); Classroom / conference room σ = 0.5 (personnel density dominates).

[0118] ΔOcc represents the personnel density change rate predicted by the LSTM model, which is 0-1.

[0119] CO2 represents the CO2 concentration of the target area.

[0120] λ represents the CO2 compensation coefficient, which is generally set to λ = 0.002 by default, and is strongly related to personnel density, which will be adjusted within a certain range with the change of personnel density.

[0121] S43: Calculate the comfort weight γ.

[0122] When the energy consumption weight α and the CO2 exceeding weight β are calculated through the above steps S41 and S42 respectively, then γ = 1-α-β. And in this embodiment, γ also needs to meet a constraint condition, γ ≥ γ min . Wherein, γ min is the preset lower limit value of comfort weight (for example 0.1), which can be configured according to different building types or user requirements.

[0123] In this embodiment, γ min represents the lower limit of the comfort weight of the target area. If γ ≥ γ min , maintain the current weight; if γ < γ min , let γ = γ min , and compress the energy consumption weight α and the CO2 exceeding weight β in proportion.

[0124] The above steps S41-S43 will be illustrated by examples as follows:

[0125] For example: the predicted personnel density increases by 30% (i.e. ΔOcc = 0.3), the outdoor temperature T out = 28℃ (T set = 24℃), and σ = 0.3, λ = 0.002, CO2 = 850ppm.

[0126] (1) Calculate α:

[0127] T out -T set = 4℃, α = 1 / 1+e-0.5×4 = 0.88.

[0128] (2) Calculate β:

[0129] β=0.3×0.3+0.002×(850-800)=0.09+0.1=0.19.

[0130] (3) Calculate γ:

[0131] γ=1-0.88-0.19=-0.07.

[0132] Since -0.07 is less than 0.1, the constraint is triggered, γ = 0.1, and α and β are redistributed based on the compression factor;

[0133] Compression factor = (1-γ min ) / (α+β)=0.9 / 1.07≈0.8411;

[0134] New α = original α Compression factor = 0.88 0.8411≈0.740;

[0135] New β = original β Compression factor = 0.19 0.8411≈0.16;

[0136] Final weights: α = 0.74, β = 0.16, γ = 0.1.

[0137] S44: Long-term output air volume of the fresh air system.

[0138] After the energy consumption weight α, CO2 excess weight β, and comfort weight γ are calculated based on the above steps S41 to S43, the long-term output air volume of the fresh air system is output through weighted calculation. Specifically, the following steps are included:

[0139] S441: Calculate the energy consumption recommended air volume based on the minimum energy consumption curve;

[0140] S442: Obtaining air quality recommended air volume based on CO2 change rate;

[0141] S443: Based on the comfort level, obtain the comfort recommended air volume required to maintain the human body thermal response PMV∈[-0.5,+0.5];

[0142] S444: Based on the maximum air volume of the fresh air system, normalize the recommended air volume for energy consumption, air quality, and comfort; where u energy = Recommended air volume for energy consumption / Maximum air volume for the system 100%,u air = Recommended air volume for air quality / Maximum air volume for the system 100%, u comfort = Comfort recommendation air volume / System maximum air volume 100%;

[0143] S445: Based on the normalized calculation result, obtain the final weighted calculation result u of the fresh air system final , ;

[0144] S446: According to the final weighted calculation result u final , calculate the long-term output air volume Q of the fresh air system final , Q final = u final System maximum air volume.

[0145] Next, the specific calculation process of this step S44 is illustrated by example:

[0146] For convenience of calculation, it is assumed that in another scenario, the weights calculated by the hierarchical DRL are: energy consumption weight α = 0.6, CO2 overweight weight β = 0.3, and comfort weight γ = 0.1, and the system maximum air volume is 1500 m³ / h. The independently calculated recommended air volumes of each strategy are as follows:

[0147] (1) Independently calculated recommended values of each strategy:

[0148] u energy Energy optimization strategy: based on minimum energy consumption curve calculation, recommended air volume 800 m³ / h (low wind speed energy saving);

[0149] u air Air quality strategy: dynamically adjusted according to CO2 change rate, recommended air volume 1200 m³ / h (fast dilution of CO2);

[0150] u comfort Comfort strategy: maintain the required air volume for PMV ∈ [-0.5, +0.5], recommended air volume 1000 m³ / h (maintain PMV = 0).

[0151] (2) Instruction normalization calculation:

[0152]

[0153] (3) Weighted synthesis calculation:

[0154]

[0155] (4) Physical quantity back calculation

[0156] .

[0157] It can be understood that the long-term output air volume is related to the three parameters of a, b, and g. Since the three parameters of a, b, and g are a dynamic adjustment process, the long-term output air volume Q of the system is also a dynamic change control result with the changes of the three parameters of a, b, and g. final

[0158] S50: Obtain an execution judgment result of the fresh air system based on a preset judgment logic.

[0159] In this embodiment, as shown in the figure, Figure 5 the preset judgment logic includes:

[0160] S51: Calculate an output air volume difference.

[0161] Specifically, the short-term output air volume of the fresh air system is obtained through the short-term control instruction of step S30, and the long-term output air volume of the fresh air system is obtained through the long-term control instruction of step S40. Therefore, based on the short-term output air volume and the long-term output air volume, the output air volume difference is calculated.

[0162] S52: Compare the output air volume difference with a maximum allowed deviation of air volume control preset by the fresh air system.

[0163] S53: Execute a weighted control instruction.

[0164] If the output air volume difference is not greater than the maximum allowed deviation of air volume control, it indicates that the air volume control is normal. At this time, the short-term output air volume and the long-term output air volume are weighted and averaged according to a preset proportion (for example, the short-term output air volume is 70% and the long-term output air volume is 30%), and the weighted control instruction is executed based on the weighted average result (i.e., the final weighted system air volume).

[0165] S54: Trigger an arbitration mechanism and execute a corresponding control instruction based on an arbitration result.

[0166] If the output air volume difference is greater than the maximum allowed deviation of air volume control, the arbitration mechanism is triggered to output an arbitration result based on the arbitration mechanism, so that a corresponding control instruction is executed based on the arbitration result. The arbitration result is to directly execute the short-term control instruction or to execute a weighted control instruction synthesized based on the short-term control instruction and the long-term control instruction after adjusting the weight.

[0167] Specifically, as shown in the figure, Figure 6 the following steps are included:

[0168] S541: Conflict detection based on the short-term control instruction and the long-term control instruction;

[0169] ​Specifically, in the embodiment, when any one of the scenarios in Table 1 occurs, the short-term control instruction is preferentially adopted. When any one of the scenarios in Table 2 occurs, the long-term control instruction is preferentially adopted. It should be understood that, in the arbitration conflict detection, if the scenario shown in Table 1 is identified, the corresponding short-term response action is preferentially executed (covering the original short-term control instruction); if the scenario shown in Table 2 is identified, the corresponding long-term strategy modification is preferentially executed (covering or modifying the original long-term control instruction).

[0170] Table 1: Scenario for preferentially adopting short-term control instruction

[0171]

[0172] Table 2: Scenario for preferentially adopting long-term control instruction

[0173]

[0174] S542: If there is a conflict, the short-term control instruction is preferentially executed;

[0175] S543: If there is no conflict, it is determined whether the CO2 is over standard; if the CO2 is over standard, the weight of the short-term control instruction for the air volume is increased by 30 percentage points when the control instruction is synthesized; if the CO2 is not over standard, the weight of the long-term control instruction for the air volume is increased by 20 percentage points, and the final execution instruction is formed. It should be understood that the "percentage point" here is to clearly indicate the absolute increase of the air volume weight, for example, the air volume weight is increased by 30% or 20% in absolute value, not relatively increased by 30% or 20%.

[0176] It can be understood that, in the embodiment, the execution judgment result obtained based on the above judgment logic is one of the short-term control instruction or the weighted control instruction of the short-term and long-term. In the embodiment, the weighted control instruction has two forms: (1) when the arbitration mechanism is not triggered, the weighted control instruction is the instruction obtained by weighted average of the short-term output air volume and the long-term output air volume according to a preset ratio; (2) when the arbitration mechanism is triggered, the weighted control instruction is the weighted control instruction synthesized again based on the adjustment of the weights of the short-term control instruction and the long-term control instruction according to the preset ratio.

[0177] S60: Fresh air system control.

[0178] Specifically, based on the execution judgment result determined in the above step S50, the air volume control result of the fresh air system is obtained. It can be understood that, if the execution judgment result is the short-term control instruction, the air volume control result is the short-term output air volume; if the execution judgment result is the weighted control instruction of the short-term and long-term, the weighted calculation result of the common output air volume is obtained through the weighted calculation of the short-term output air volume and the long-term output air volume.

[0179] When the required output air volume is determined, the final air volume of the fresh air system is equal to the required output air volume by controlling the fan speed, valve opening and bypass valve opening of the fresh air system.

[0180] In addition, in a preferred embodiment of the present application, the energy-saving control method further comprises:

[0181] S70: Abnormality processing.

[0182] Specifically, first, for the CO2 sensor array and the external environment sensor, the Mahalanobis distance is used to detect the abnormal value of the sensor:

[0183] ;

[0184] Wherein, D M represents the Mahalanobis distance value, which quantifies the deviation of the current observation value x from the normal state defined by μ and Σ; x represents the observation value vector of the current sensor; μ represents the mean vector of the sensor data; represents the deviation of the current data from the normal state. represents the inverse matrix of the covariance matrix.

[0185] Secondly, based on the preset threshold and the abnormal value of each sensor, it is judged whether each sensor is failed.

[0186] Finally, if a single external environment sensor fails, switch to the redundant sensor; if the sensor array fails, enable the GAN data generator to replace the sensor array.

[0187] Therefore, by detecting the failure of a single sensor or a sensor array and taking corresponding remedial measures, the fault tolerance of the energy-saving control method is improved, and the stability of the fresh air system operation can be improved.

[0188] Second embodiment

[0189] As Figure 7 shown, based on the first embodiment described above, the second embodiment of the present application provides an energy-saving control device for a fresh air system, which comprises a data acquisition layer 1, an edge computing layer 2 and an execution control layer 3. Among them,

[0190] The data acquisition layer 1 comprises a CO2 sensor array, an infrared detection device, a Wi-Fi detection device, an enthalpy difference calculation module and external environment sensors and other components to obtain the current environmental detection data of the target area.

[0191] Edge computing layer 2 has a built-in LSTM model and a hierarchical DRL controller and is connected to data acquisition layer 1. This layer is used to predict the rate of change of occupant density in the short term based on the LSTM model to output short-term control instructions. Furthermore, this layer is used to output long-term control instructions based on the hierarchical DRL controller.

[0192] The execution control layer 3 is connected to the edge computing layer 2 and has a preset judgment logic to obtain the execution judgment result of the fresh air system based on the preset judgment logic; and based on the execution judgment result, the air volume control result of the fresh air system is obtained to control the fan speed, valve opening and bypass valve opening of the fresh air system based on the air volume control result.

[0193] It can be understood that the functions and connection relationships of the various module units in this embodiment are only a specific implementation method for realizing the energy-saving control method in the above-mentioned first embodiment. In other embodiments, the functions and connection relationships of the various module units can be adaptively adjusted as needed to realize the energy-saving control method in the above-mentioned first embodiment, and no specific limitation is made here.

[0194] Third embodiment

[0195] like Figure 8 As shown, based on the above energy-saving control method for a fresh air system, the third embodiment of the present invention further provides another energy-saving control device for a fresh air system. The energy-saving control device includes one or more processors and a memory. The memory is coupled to the processor and is used to store one or more programs. When the programs are executed by the processor, the processor implements the energy-saving control method of the above embodiment.

[0196] The processor is used to control the overall operation of the energy-saving control device to complete all or part of the steps of the above-mentioned energy-saving control method. The processor can be a central processing unit (CPU), a graphics processing unit (GPU), a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), a digital signal processing (DSP) chip, etc. The memory is used to store various types of data to support the operation of the energy-saving control device. This data may include, for example, instructions for any application or method operating on the energy-saving control device, as well as application-related data. The memory can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, etc.

[0197] In one example embodiment, the energy-saving control device can be implemented by a computer chip or entity, or by a product with certain functions, for implementing the energy-saving control method described above and achieving the technical effects consistent with the above method. A typical embodiment is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, an in-vehicle human-machine interaction device, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0198] In another example embodiment, the present application also provides a computer-readable storage medium including program instructions, which, when executed by a processor, implement the steps of the energy-saving control method in any one of the above embodiments. For example, the computer-readable storage medium can be the memory described above including program instructions, which can be executed by the processor of the energy-saving control device to complete the energy-saving control method described above and achieve the technical effects consistent with the above method.

[0199] It should be noted that the above embodiments are only illustrative. The technical solutions of various embodiments can be combined, and all are within the protection scope of the present application.

[0200] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise explicitly specified.

[0201] The energy-saving control method and device for fresh air systems provided by the present application are described in detail above. For those skilled in the art, any obvious modification made to it without departing from the essential content of the present application will constitute an infringement of the patent right of the present application and will bear the corresponding legal responsibility.

Claims

1. An energy-saving control method for a fresh air system, characterized in that The steps include: Obtain the current environmental detection data of the target area; Combining the current environmental monitoring data with the historical environmental monitoring data of the target area and inputting the data into a preset LSTM model to predict the rate of change of population density in the short term in the future; wherein the LSTM model is trained based on the historical environmental monitoring data of the target area; Based on the population density change rate in the short term in the future, output a short-term control instruction; wherein the short-term control instruction is used to directly control the short-term output air volume of the fresh air system; The population density change rate in the short term in the future is combined with the current environmental detection data and inputted into the hierarchical DRL controller to output a long-term control instruction; wherein the long-term control instruction is used to adjust the energy consumption weight, CO2 excess weight, and comfort weight of the fresh air system to control the long-term output air volume of the fresh air system by weighted average; Based on the preset judgment logic, obtaining the execution judgment result of the fresh air system; wherein the execution judgment result is one of a short-term control instruction or a short-term and long-term weighted control instruction; Based on the execution judgment result, the air volume control result of the fresh air system is obtained, so as to control the fan speed, valve opening and bypass valve opening of the fresh air system based on the air volume control result.

2. The energy-saving control method according to claim 1, characterized in that The obtaining of the current environmental detection data of the target area specifically includes: The CO2 concentration in the target area is obtained by deploying a dual CO2 sensor array on the ceiling of the target area; Use infrared detection devices in conjunction with Wi-Fi detection devices to detect the current population density in the target area; The enthalpy difference value of the target area is obtained by deploying an enthalpy difference calculation module in the corridor of the target area; Obtain outdoor temperature, humidity, and PM2.5 test results in the target area through external environmental sensors; Among them, the CO2 concentration, current population density, enthalpy difference, outdoor temperature and humidity, and PM2.5 detection results of the target area together constitute the current environmental detection data of the target area.

3. The energy-saving control method according to claim 2, characterized in that The outputting of short-term control instructions based on the population density change rate in the short term in the future specifically includes: If the rate of change of the population density is greater than a first threshold, the short-term control instruction is to rapidly increase the air volume and close the bypass valve; If the rate of change of the population density is not greater than the first threshold and greater than the second threshold, the short-term control instruction is to increase the air volume in a step-by-step manner by increasing the air volume gradient by one per preset interval; If the rate of change of the personnel density is not greater than the second threshold and greater than the third threshold, the short-term control instruction is to maintain the current air volume and fine-tune the air volume by a preset amplitude; If the rate of change of the personnel density is not greater than a third threshold, the short-term control instruction is to reduce the air volume and start the night cleaning mode; The first threshold value>the second threshold value>the third threshold value, and the first threshold value and the second threshold value are both positive percentages, and the third threshold value is a negative percentage.

4. The energy-saving control method according to claim 2, characterized in that The long-term control instruction is obtained by the following steps: Based on the current environmental detection data, the outdoor temperature Tout is obtained; and the energy consumption weight α is calculated based on the temperature setting value Tset and the temperature sensitivity coefficient k. ; Based on the population density change rate and the CO2 concentration in the target area, the CO2 excess weight β is calculated. ; Wherein, σ represents the population density sensitivity coefficient; ΔOcc represents the population density change rate predicted by the LSTM model, ranging from 0 to 1; CO2 represents the CO2 concentration in the target area; λ represents the CO2 compensation coefficient; and 800 represents the target threshold value of CO2 concentration. Based on the energy consumption weight α and the CO2 exceeding standard weight β, the comfort weight γ is calculated, γ=1-α-β and γ≥γ min ; Among them, γ min represents the lower limit of the comfort weight of the target area; if γ≥γ min , then maintain the current weight; if γ<γ min , then let γ=γ min , and proportionally compress the energy consumption weight α and the CO2 exceeding standard weight β; Based on the energy consumption weight α, the CO2 exceeding standard weight β and the comfort weight γ, the long-term output air volume of the fresh air system is output through weighted calculation.

5. The energy-saving control method according to claim 4, characterized in that The long-term output air volume is weightedly calculated by the following steps: Calculate the energy consumption recommended air volume based on the minimum energy consumption curve; Get recommended air volume for air quality based on CO2 change rate; According to the comfort level, obtain the recommended air volume required to maintain the human body thermal response PMV∈[-0.5,+0.5]; Based on the maximum air volume of the fresh air system, the energy consumption recommended air volume, air quality recommended air volume and comfort recommended air volume are normalized; wherein, u energy = Recommended air volume for energy consumption / Maximum air volume for the system 100%,u air = Recommended air volume for air quality / Maximum air volume for the system 100%,u comfort = Recommended air volume for comfort / Maximum air volume of the system 100%; Based on the normalized calculation results, the final weighted calculation result u of the fresh air system is obtained final ,in ; According to the final weighted calculation result u final , calculate the long-term output air volume Q of the fresh air system final , where Q final =u final Maximum air volume of the system.

6. The energy-saving control method according to claim 1, characterized in that The preset judgment logic specifically includes: Calculating an output air volume difference based on the short-term output air volume and the long-term output air volume; Comparing the output air volume difference with the maximum allowable deviation of air volume control preset for the fresh air system; If the output air volume difference is not greater than the maximum allowable deviation of air volume control, a weighted average is performed based on the short-term output air volume and the long-term output air volume according to a preset ratio, so as to execute a weighted control instruction based on the weighted average result; If the output air volume difference is greater than the maximum allowable deviation of the air volume control, the arbitration mechanism is triggered to output an arbitration result based on the arbitration mechanism, and thus the corresponding control instruction is executed based on the arbitration result; wherein the arbitration result is to directly execute the short-term control instruction, or to execute a weighted control instruction synthesized after adjusting the weights of the short-term control instruction and the long-term control instruction.

7. The energy-saving control method according to claim 6, characterized in that The arbitration mechanism specifically includes: performing conflict detection based on the short-term control instruction and the long-term control instruction; If there is a conflict, the short-term control instruction shall be executed first; If there is no conflict, continue to determine whether CO2 exceeds the standard; if CO2 exceeds the standard, increase the weight of the short-term control instruction by 30%; if CO2 does not exceed the standard, increase the weight of the long-term control instruction by 20% as the weighted control instruction after weight adjustment.

8. The energy-saving control method according to claim 2, characterized in that Also includes: For the CO2 sensor array and external environment sensor, Mahalanobis distance is used to detect sensor abnormal values: ； Among them, D M Represents the Mahalanobis distance value, x represents the observation value vector of the current sensor, μ represents the mean vector of the sensor data, Indicates the deviation of the current data from the normal state, represents the inverse matrix of the covariance matrix; Based on the preset threshold and the abnormal value of each sensor, determine whether each sensor is invalid; If a single external environment sensor fails, the redundant sensor is switched; if the sensor array fails, the GAN data generator is enabled to replace the sensor array.

9. An energy-saving control device for a fresh air system, used to implement the energy-saving control method according to any one of claims 1 to 8, characterized in that include: The data collection layer is used to obtain the current environmental detection data of the target area; An edge computing layer, having a built-in preset LSTM model and a hierarchical DRL controller, and connected to the data acquisition layer; the edge computing layer is used to predict the rate of change of population density in the short term in the future based on the LSTM model to output short-term control instructions; the edge computing layer is also used to output long-term control instructions based on the hierarchical DRL controller; An execution control layer is connected to the edge computing layer and has a preset judgment logic to obtain an execution judgment result of the fresh air system based on the preset judgment logic; and based on the execution judgment result, an air volume control result of the fresh air system is obtained to control the fan speed, valve opening and bypass valve opening of the fresh air system based on the air volume control result.

10. An energy-saving control device for a fresh air system, characterized in that The system comprises a processor and a memory, wherein the processor reads a computer program in the memory to implement the energy-saving control method according to any one of claims 1 to 8.

Images