Linkage control method for energy-saving ventilation equipment

Abstract

The invention relates to the technical field of intelligent control, and particularly discloses a linkage control method for energy-saving ventilation equipment, which comprises the following steps of: acquiring indoor and outdoor temperature, indoor carbon dioxide concentration and personnel occupancy state data in real time; according to the method, deep collaborative analysis is carried out on the indoor personnel occupancy state and the deviation between the carbon dioxide concentration and a target set value, so that joint characterization of the current indoor fresh air volume regulation and control requirement is achieved, control parameters of a fresh air volume PI D controller are dynamically optimized accordingly, and a basic fresh air volume regulation instruction is calculated. On the basis, thermal disturbance introduced by fresh air is further quantified based on the outdoor and indoor temperature difference, and the thermal disturbance serves as a feedforward signal to be compensated to an air conditioner load adjusting loop, so that dynamic adjustment of the air conditioner load is achieved. In this way, cooperative control over ventilation and air conditioning systems can be achieved, the fresh air-air conditioning equipment energy consumption coupling conflict in traditional control is restrained, and therefore the building energy consumption is reduced while the indoor environment quality is guaranteed.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology, and more specifically, to a linkage control method for energy-saving ventilation equipment. Background Art

[0002] In modern building complexes, maintaining appropriate indoor air quality (IAQ) and thermal comfort are the core objectives of the heating, ventilation, and air conditioning (HVAC) system. Among them, the ventilation system is responsible for introducing fresh outdoor air, diluting and exhausting indoor pollutants (such as carbon dioxide, formaldehyde, etc.), and ensuring the health of occupants; the air conditioning system is responsible for regulating indoor temperature and humidity to create a comfortable thermal environment. However, there are inherent couplings and conflicts in the operation of these two systems: the temperature and humidity of the outdoor fresh air introduced by the ventilation system often differ from the indoor set values, which will directly form additional cooling / heating loads and increase the energy consumption of the air conditioning system; conversely, the operation strategy of the air conditioning system may also affect indoor environmental parameters, thereby changing the demand for fresh air volume. If the two are controlled independently without coordination, it often leads to a disconnection between the fresh air intake and the actual demand (resulting in excessive ventilation or insufficient ventilation), or the air conditioning system frequently adjusts to cope with the disturbance caused by fresh air, resulting in unnecessary energy waste, and may also affect the stability and comfort of the indoor environment. Therefore, in order to minimize building energy consumption while meeting indoor environmental requirements, developing an energy-saving linkage control method for ventilation equipment to achieve efficient coordinated operation of the ventilation and air conditioning systems has become an important research direction and urgent need in the HVAC field.

[0003] Currently, there have been some attempts to conduct linkage control on the ventilation and air conditioning systems. For example, some systems adopt simple time-division control or start-stop strategies based on fixed thresholds to adjust the opening of the fresh air valve and the start and stop of the air conditioner according to a preset schedule or a single environmental parameter. On the one hand, the existing control methods do not adequately consider the coupling effect between systems, especially the immediate disturbance caused by fresh air introduction to indoor temperature, resulting in a lag in the response of the air conditioning system. On the other hand, when adjusting the fresh air volume, most use controllers with fixed parameters (such as PID controllers), and their control parameters (proportional, integral, and derivative gains) are usually set according to typical working conditions during the commissioning stage, making it difficult to adapt to the complex non-linear characteristics brought about by the dynamic changes of various factors such as indoor and outdoor environments and occupancy status during actual operation. When the working conditions deviate from the design point, the fixed-parameter controller may exhibit problems such as slow response, excessive overshoot, or insufficient stability, unable to accurately and efficiently match the real-time fresh air demand, affecting the control accuracy and energy-saving effect. Especially in large public buildings, the spatio-temporal differences in the occupancy density make it difficult for the fixed-threshold control parameters to adapt to the dynamic environment, easily leading to prominent problems such as frequent system oscillations and adjustment lags.

[0004] Therefore, an optimized interlocking control method for energy-saving ventilation equipment is expected. Summary of the Invention

[0005] To solve the above technical problems, the present application is proposed. An embodiment of the present application provides an interlocking control method for energy-saving ventilation equipment, which collects real-time indoor and outdoor temperature, indoor carbon dioxide concentration and occupancy status data, and through in-depth collaborative analysis of the indoor occupancy status and the deviation between the carbon dioxide concentration and the target set value, realizes the joint characterization of the current indoor fresh air volume regulation demand, and accordingly dynamically optimizes the control parameters of the fresh air volume PID controller to calculate the basic fresh air volume adjustment instruction. On this basis, based on the temperature difference between the outdoor and indoor, the heat disturbance introduced by the fresh air is further quantified and used as a feedforward signal to compensate the air-conditioning load regulation loop to realize the dynamic adjustment of the air-conditioning load. In this way, the coordinated control of the ventilation and air-conditioning systems can be achieved, and the energy consumption coupling conflict between the fresh air and air-conditioning equipment in traditional control can be suppressed, thereby reducing the building energy consumption while ensuring the indoor environmental quality.

[0006] Correspondingly, according to one aspect of the present application, there is provided an interlocking control method for energy-saving ventilation equipment, which includes:

[0007] Obtain the real-time indoor carbon dioxide concentration value and the real-time indoor temperature value;

[0008] Obtain the outdoor temperature value and the indoor occupancy status data;

[0009] Based on the deviation between the real-time indoor carbon dioxide concentration value and the carbon dioxide target value and the indoor occupancy status data, calculate the basic fresh air volume adjustment instruction;

[0010] Based on the deviation between the real-time indoor temperature value and the temperature target value and the indoor occupancy status data, calculate the basic air-conditioning load adjustment instruction;

[0011] Based on the basic fresh air volume adjustment instruction and the difference between the outdoor temperature value and the real-time indoor temperature value, estimate the disturbance load introduced by the fresh air volume;

[0012] Input the disturbance load introduced by the fresh air volume as a feedforward signal into the temperature control loop to obtain the final air-conditioning load adjustment instruction.

[0013] Compared with the prior art, the linkage control method for energy-saving ventilation equipment provided by the present application collects real-time indoor and outdoor temperature, indoor carbon dioxide concentration, and personnel occupancy status data. By deeply synergistically analyzing the deviation between the indoor personnel occupancy status and the carbon dioxide concentration and the target set value, it realizes the joint representation of the current indoor fresh air volume regulation demand, and accordingly dynamically optimizes the control parameters of the fresh air volume PID controller to calculate the basic fresh air volume adjustment instruction. On this basis, based on the temperature difference between the outdoor and indoor, the heat disturbance introduced by fresh air is further quantified and used as a feedforward signal to compensate the air-conditioning load regulation loop to achieve dynamic adjustment of the air-conditioning load. In this way, the coordinated control of the ventilation and air-conditioning systems can be realized, suppressing the energy consumption coupling conflict of the fresh air-air-conditioning equipment in traditional control, thereby reducing building energy consumption while ensuring the indoor environmental quality. Description of the Drawings

[0014] By describing the embodiments of the present application in more detail in combination with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0015] Figure 1 It is a flowchart of the linkage control method for energy-saving ventilation equipment according to an embodiment of the present application.

[0016] Figure 2 It is a flowchart of step S3 in the linkage control method for energy-saving ventilation equipment according to an embodiment of the present application.

[0017] Figure 3 It is a flowchart of step S32 in the linkage control method for energy-saving ventilation equipment according to an embodiment of the present application.

[0018] Figure 4 It is a schematic diagram of data flow in step S32 in the linkage control method for energy-saving ventilation equipment according to an embodiment of the present application.

[0019] Figure 5 It is a flowchart of step S323 in the linkage control method for energy-saving ventilation equipment according to an embodiment of the present application. Detailed Description of the Embodiments

[0020] Next, exemplary embodiments according to the present application will be described in detail with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0021] Figure 1 This is a flowchart of a linkage control method for an energy-saving ventilation device according to an embodiment of the present application. As Figure 1 shown, the linkage control method for an energy-saving ventilation device according to an embodiment of the present application includes the steps of: S1, obtaining the real-time indoor carbon dioxide concentration value and the real-time indoor temperature value; S2, obtaining the outdoor temperature value and the indoor occupancy status data; S3, calculating a basic fresh air volume adjustment instruction based on the deviation between the real-time indoor carbon dioxide concentration value and the carbon dioxide target value and the indoor occupancy status data; S4, calculating a basic air-conditioning load adjustment instruction based on the deviation between the real-time indoor temperature value and the temperature target value and the indoor occupancy status data; S5, estimating the disturbance load introduced by the fresh air volume based on the basic fresh air volume adjustment instruction and the difference between the outdoor temperature value and the real-time indoor temperature value; S6, inputting the disturbance load introduced by the fresh air volume as a feedforward signal into the temperature control loop to obtain a final air-conditioning load adjustment instruction.

[0022] In the above-mentioned linkage control method for energy-saving ventilation equipment, in step S1, the real-time indoor carbon dioxide concentration value and the real-time indoor temperature value are obtained. It should be understood that the carbon dioxide concentration is a key indicator for measuring indoor air quality (IAQ), which is directly related to personnel activities and fresh air demand, while the indoor temperature is the core parameter for evaluating thermal comfort and determines the load demand of the air conditioning system. Therefore, in order to provide real-time and accurate feedback input for subsequent fresh air volume adjustment and air conditioning load adjustment, and ensure that the control strategy can truly reflect the actual situation of the indoor environment. This application is based on the basic principle of closed-loop feedback control, that is, the generation of the control action must be based on the actual measured value of the controlled quantity. By deploying high-precision carbon dioxide sensors (such as non-dispersive infrared NDIR sensors) and temperature sensors (such as thermistors, RTDs or integrated digital temperature and humidity sensors) in representative indoor areas (such as near the return air outlet or areas with intensive personnel activities), and through wired (such as 4-20mA analog signals, RS485 bus using Modbus protocol) or wireless (such as Zigbee, LoRaWAN) methods, the real-time concentration values (usually in ppm) and temperature values (in °C) collected by these sensors continuously or at a high enough frequency (such as every minute or every few minutes) are transmitted to the data acquisition interface of the central controller (such as PLC, DDC or embedded control system) to achieve real-time data processing and decision-making. Specifically, after receiving the electrical signal or digital message from the sensor, the controller will perform necessary signal conditioning (such as filtering, amplification), analog-to-digital conversion (if it is an analog signal) and unit conversion, and finally obtain the standardized and digital real-time indoor carbon dioxide concentration value and real-time temperature value available for the control algorithm, and store them in the corresponding memory variables for subsequent steps to call. In this way, the dynamic changes of indoor air quality and thermal environment can be grasped in real time, laying a solid data foundation for subsequent refined and linked control based on actual needs.

[0023] In the above-mentioned linkage control method for energy-saving ventilation equipment, in step S2, the outdoor temperature value and indoor occupancy status data are obtained. It should be understood that since the temperature of the outdoor fresh air introduced by the ventilation system directly affects the indoor heat load, it is a key parameter for calculating the fresh air disturbance and achieving feedforward compensation. At the same time, whether there are people indoors and the personnel density are the fundamental factors determining whether ventilation and air conditioning need to operate and the operation intensity, which are directly related to the magnitude of energy-saving potential. Therefore, in order to enable the control system to evaluate the energy impact brought by the introduction of fresh air and adjust the control objectives and strategies according to the actual usage requirements. This application is based on the ideas of predictive control and demand-side management, that is, the control decision should not only consider the current indoor state, but also incorporate the key scenario information of external environmental impacts and internal actual needs. By installing an outdoor temperature sensor at a suitable outdoor location (usually in a shaded and well-ventilated place, avoiding direct sunlight and the influence of local heat sources), the accurate outdoor air temperature is obtained. At the same time, one or more technical means are used to judge the indoor occupancy status, such as deploying passive infrared (PIR) motion sensors to detect personnel activities, using cameras for people counting, integrating access control system card swiping records, or allowing users to manually set the occupancy mode through a panel, etc., so as to obtain the real-time indoor occupancy status data (including occupied / unoccupied, number of people, personnel density, activity mode). In this way, the key external parameter (outdoor temperature) for calculating the fresh air heat load and the core internal demand information (occupancy status) determining the control mode and energy consumption level are effectively obtained, enabling the subsequent formulation of control instructions to fully consider energy-saving constraints and actual needs, avoiding wasting energy during idle periods, and providing the necessary conditions for accurately calculating the fresh air disturbance load.

[0024] In the above-mentioned linkage control method for energy-saving ventilation equipment, in step S3, based on the deviation between the indoor real-time carbon dioxide concentration value and the carbon dioxide target value and the indoor occupancy status data, a basic fresh air volume adjustment instruction is calculated. It should be understood that the fresh air volume demand is closely related to indoor personnel activities and carbon dioxide concentration. When the indoor carbon dioxide concentration is higher than the target value, it indicates that the air quality has deteriorated and more fresh air needs to be introduced to dilute indoor pollutants. At the same time, the indoor occupancy status (such as the number of people, activity level) also affects the fresh air demand. When people are dense or activities are frequent, the fresh air demand increases accordingly. Therefore, in this application, by comparing the indoor real-time carbon dioxide concentration value with the preset carbon dioxide target value, the concentration deviation is calculated, and combined with the indoor occupancy status data (such as whether there are people, the number of people, etc.), an appropriate control algorithm is used to convert the concentration deviation and occupancy status into specific demands for the fresh air volume, and a basic fresh air volume adjustment instruction (the introduced fresh air mass flow value) is obtained to reveal the fresh air volume required based on the current actual indoor conditions while maintaining the indoor air quality up to standard.

[0025] Figure 2It is a flowchart of step S3 in the linkage control method for an energy-saving ventilation device according to an embodiment of the present application. As Figure 2 shown, the step S3 includes: S31, calculating the deviation between the indoor real-time carbon dioxide concentration value and the carbon dioxide target value to obtain the carbon dioxide concentration target offset; S32, adaptively optimizing the fresh air volume control parameters based on the deep collaborative association response characteristics between the carbon dioxide concentration target offset and the indoor occupancy status data to obtain the controller optimization parameters; S33, updating the controller optimization parameters to the parameter register of the fresh air volume PID controller to obtain an optimized fresh air volume PID controller; S34, inputting the carbon dioxide concentration target offset into the optimized fresh air volume PID controller to obtain the basic fresh air volume adjustment instruction.

[0026] Specifically, in the step S31, the deviation between the indoor real-time carbon dioxide concentration value and the carbon dioxide target value is calculated to obtain the carbon dioxide concentration target offset. That is, in order to quantify the dynamic deviation state of the current indoor air pollution degree from the target value, the present application constructs a quantitative representation of the pollution load through real-time difference calculation based on the deviation feedback control principle. Specifically, an embedded system can be used to read the real-time concentration value (such as 1200 ppm) collected by a carbon dioxide sensor and perform a subtraction operation with a preset health standard target value (such as 800 ppm) to obtain the current concentration offset (+400 ppm). This offset contains both absolute deviation information and the pollutant generation rate characteristics caused by personnel activities. In this way, the degree of air quality deterioration can be converted into an engineering variable that can be quantitatively controlled, providing an accurate input reference for subsequent dynamic adjustment.

[0027] Specifically, in the step S32, the fresh air volume control parameters are adaptively optimized based on the deep collaborative association response characteristics between the carbon dioxide concentration target offset and the indoor occupancy status data to obtain the controller optimization parameters. Among them, Figure 3 It is a flowchart of step S32 in the linkage control method for an energy-saving ventilation device according to an embodiment of the present application. Figure 4 It is a data flow diagram of step S32 in the linkage control method for an energy-saving ventilation device according to an embodiment of the present application. As Figure 3 and Figure 4As shown, step S32 includes: S321, performing one-hot encoding on the carbon dioxide concentration target offset to obtain a one-hot embedded encoding vector of the carbon dioxide concentration target offset; S322, performing structured embedding encoding on the indoor occupancy state data to obtain a structured embedded encoding vector of the indoor occupancy state; S323, performing fine-grained collaborative response correlation analysis on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy state to obtain a target offset-dynamic gain fine-grained collaborative encoding vector; S324, performing feature decoding on the target offset-dynamic gain fine-grained collaborative encoding vector to obtain controller optimization parameters, where the controller optimization parameters include an optimization ratio parameter, an optimization integral parameter, and an optimization differential parameter.

[0028] Specifically, in step S321, one-hot encoding is performed on the carbon dioxide concentration target offset to obtain a one-hot embedded encoding vector of the carbon dioxide concentration target offset. It should be understood that discrete concentration deviation values are difficult to directly represent the control priorities of different pollution levels, and the subsequent complex feature analysis network lacks the adaptability to discrete state features and cannot fully highlight the different control response requirements implied by different "intervals" of offsets (for example, small offsets, medium offsets, and severe over-limit may require very different control strategies). Therefore, in order to map continuous deviations into discrete state features with semantic stratification, based on the classification feature encoding principle, this application constructs a hierarchical representation of pollution levels through one-hot encoding technology. Specifically, first, a deviation interval division rule is set (for example: ±50 ppm is "normal", +50 to +200 ppm is "mild pollution", +200 to +500 ppm is "moderate pollution", and above +500 ppm is "severe pollution"). Then, the real-time carbon dioxide concentration target offset is converted into a corresponding one-hot encoding vector according to the preset interval division rule. For example, if the real-time offset is +300 ppm, according to the rule, this offset falls into the "moderate pollution" interval, and the corresponding one-hot encoding vector is [0, 0, 1, 0], where "1" indicates that the current state is "moderate pollution", and the remaining bits being "0" indicate non-current states. In this way, it helps to convert fuzzy concentration deviations into a feature space with clear control directions, providing a structured input for subsequent dynamic parameter optimization.

[0029] Specifically, in step S322, the indoor occupancy state data is structured embedded and coded to obtain an indoor occupancy state structured embedded coding vector. It should be understood that the indoor occupancy density directly affects the spatial differences in the carbon dioxide generation rate and the fresh air demand, which is another important basis for dynamically adjusting the indoor environmental parameters. However, the original occupancy state data (including occupancy / unoccupancy, number of people, occupancy density, and activity mode) are expressed as non-numerical text descriptions, which are difficult to directly interact with the carbon dioxide concentration offset characteristics. Therefore, in order to integrate the occupancy state characteristics into the control decision process and realize the comprehensive analysis of multi-dimensional features, the present application uses structured embedded coding technology to deeply process the indoor occupancy state data. In an embodiment of the present application, a natural language processing technology based on BERT is used to convert the original occupancy state data (such as "someone", "no one", "3 people", "highly dense", etc.) into a high-dimensional dense vector with context relevance, and obtain an indoor occupancy state structured embedded coding vector, which provides a more detailed and comprehensive occupancy feature description for the dynamic adjustment of the fresh air volume.

[0030] Specifically, the step S323 performs fine-grained collaborative response association analysis on the one-hot embedded coding vector of the target offset of the carbon dioxide concentration and the structured embedded coding vector of the indoor occupancy state to obtain a target offset-dynamic gain fine-grained collaborative coding vector. It should be understood that the fresh air control strategy does not simply depend on one of the carbon dioxide offset or the occupancy state, but strongly depends on the dynamic interaction between the two; for example, the same +200ppm carbon dioxide offset may require a faster and stronger response (i.e., different PID gains) when it occurs in "high-density occupancy" than in "low-density occupancy". Traditional methods find it difficult to capture this subtle and nonlinear synergistic effect. Therefore, in order to deeply explore and understand the dynamic response characteristics that the control system should have under a specific combination of carbon dioxide offset levels and specific occupancy modes (for example, whether fast response, stable transition or overshoot suppression is required), so as to provide precise guidance for adaptive adjustment of controller parameters, the present application performs a fine-grained collaborative response association analysis on the carbon dioxide concentration target offset one-hot embedded coding vector and the indoor occupancy state structured embedded coding vector to explore the optimal dynamic behavior mode of the control system indicated by the joint action of the two, and generates a target offset-dynamic gain fine-grained collaborative coding vector as a direct basis for dynamically adjusting the gain parameters of the fresh air control system.

[0031] Figure 5 FIG. 4 is a flow chart of step S323 in the linkage control method for energy-saving ventilation equipment according to an embodiment of the present application. Figure 5As shown in the figure, it includes: S3231, performing feature correlation response inference based on local granularity on the one-hot embedding encoding vector of the carbon dioxide concentration target offset and the structured embedding encoding vector of the indoor occupancy state to obtain a sequence of target offset-dynamic gain local correlation response encoding matrices; S3232, performing sequence transfer encoding on the sequence of target offset-dynamic gain local correlation response encoding matrices to obtain the target offset-dynamic gain fine-grained collaborative encoding vector.

[0032] In a specific example of the present application, the step S3231 includes: First, performing equal-granularity feature segmentation on the one-hot embedding encoding vector of the carbon dioxide concentration target offset and the structured embedding encoding vector of the indoor occupancy state to obtain a sequence of local embedding feature encoding vectors of the carbon dioxide concentration target offset and a sequence of local embedding feature encoding vectors of the indoor occupancy state, which is expressed by the formula:

[0033] Split(V1) = {x1, x2,..., x i ,..., x n}

[0034] Split(V2) = {y1, y2,..., y i ,..., y n}

[0035] Wherein, V1 represents the one-hot embedding encoding vector of the carbon dioxide concentration target offset, V2 represents the structured embedding encoding vector of the indoor occupancy state, x1, x2, x i and x n respectively represent the 1st, 2nd, ith, and nth local embedding feature encoding vectors of the carbon dioxide concentration target offset in the sequence of local embedding feature encoding vectors of the carbon dioxide concentration target offset, n is the number of local embedding feature encoding vectors of the carbon dioxide concentration target offset, y1, y2, y i and y n respectively represent the 1st, 2nd, ith, and nth local embedding feature encoding vectors of the indoor occupancy state in the sequence of local embedding feature encoding vectors of the indoor occupancy state, and Split(·) represents the feature segmentation function.

[0036] Here, since the global feature interaction analysis is directly performed on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy state, it may be difficult to capture the finer and more local association patterns between specific offset intervals and specific occupancy state dimensions. Therefore, in order to deconstruct the spatial action mechanism of high-dimensional features and achieve a fine-grained examination of the interaction between the two, this application is based on a feature divide-and-conquer strategy. By equally granular slicing, the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy state are mapped from the global encoding vector into locally analyzable local feature segments, so as to decompose the macroscopic association problem into multiple local association problems at the microscopic level, laying a foundation for subsequent more in-depth association interaction response analysis and balancing the computational efficiency.

[0037] Next, each pair of corresponding carbon dioxide concentration target offset local embedded feature encoding vectors and indoor occupancy state local embedded feature encoding vectors in the sequence of the carbon dioxide concentration target offset local embedded feature encoding vectors and the sequence of the indoor occupancy state local embedded feature encoding vectors are input into the transfer response inference unit to obtain a sequence of the target offset-dynamic gain local association response encoding matrices, which is expressed by the formula:

[0038]

[0039] Among them, W i represents the linear transformation matrix, ReLU(·) represents the ReLU activation function, represents matrix multiplication, M i represents x i and y i is the target offset-dynamic gain local association response encoding matrix between them.

[0040] That is, in order to deeply model and quantify the non-linear interaction relationship between a specific local interval of the carbon dioxide concentration target offset and the corresponding local feature dimension of the occupancy state, and capture the transfer response mechanism between the two. This application further uses a deep learning algorithm to perform transfer calculations on each pair of corresponding carbon dioxide concentration target offset local embedded feature encoding vectors and indoor occupancy state local embedded feature encoding vectors, and introduces an activation function to simulate and learn the complex non-linear relationship between the two, so as to explore the potential association and transfer trend between the indoor occupancy state and the carbon dioxide concentration target offset feature, and generate the corresponding target offset-dynamic gain local association response encoding matrix.

[0041] In a specific example of the present application, the step S3232 includes: inputting the sequence of the target offset-dynamic gain local association response encoding matrices into an LSTM transfer encoding module with an attention mechanism to obtain the target offset-dynamic gain fine-grained collaborative encoding vector. More specifically, first, flatten each target offset-dynamic gain local association response encoding matrix in the sequence of the target offset-dynamic gain local association response encoding matrices to obtain a sequence of target offset-dynamic gain local association response encoding vectors, which is expressed by the formula:

[0042] vec(M i )=v i

[0043] where vec(·) represents the matrix flattening operation, and v i represents the target offset-dynamic gain local association response encoding vector obtained by expanding M i .

[0044] Next, taking the Frobenius norm of each target offset-dynamic gain local association response encoding matrix in the sequence of the target offset-dynamic gain local association response encoding matrices as a constraint factor, perform feature modulation on the sequence of the target offset-dynamic gain local association response encoding vectors to obtain a sequence of modulated target offset-dynamic gain local association response encoding vectors, which is expressed by the formula:

[0045]

[0046] where exp(·) represents the exponential function operation with base e, is the feature modulation function based on the attention mechanism.

[0047] Furthermore, perform transfer encoding based on the LSTM model on the sequence of the modulated target offset-dynamic gain local association response encoding vectors to obtain the target offset-dynamic gain fine-grained collaborative encoding vector, which is expressed by the formula:

[0048]

[0049] where M1 represents the target offset-dynamic gain local association response encoding matrix between x1 and y1, and M n represents the target offset-dynamic gain local association response encoding matrix between x n and y n , LSTM(·) represents the LSTM model, and v f represents the target offset-dynamic gain fine-grained collaborative encoding vector.

[0050] It should be understood that the scattered target offset-dynamic gain local association response features are difficult to effectively integrate and reflect the global collaborative response characteristics. Therefore, in order to integrate the sequence of the target offset-dynamic gain local association response coding matrices into a globally consistent collaborative feature representation, the present application uses an LSTM (Long Short-Term Memory Network) transfer coding module with an attention mechanism to perform transfer coding and dynamic integration on the sequence of the target offset-dynamic gain local association response coding matrices. Specifically, this module calculates the importance weights based on the F-norm of each target offset-dynamic gain local association response coding matrix, and dynamically adjusts the contribution degrees of different local association features during the integration process through the attention mechanism, so as to emphasize the key features and suppress the non-key features. At the same time, by introducing an LSTM model to capture and remember the long-term dependence relationships in the sequence, the scattered target offset-dynamic gain local association response features can be effectively integrated to form a globally consistent collaborative feature representation, which is the target offset-dynamic gain fine-grained collaborative coding vector.

[0051] In a preferred example of the present application, the step S3232 includes: performing sparse regularization constraints on each target offset-dynamic gain local association response coding matrix in the sequence of the target offset-dynamic gain local association response coding matrices based on the transition response connection probability between the corresponding local embedding feature coding vectors of the carbon dioxide concentration target offset amount and the local embedding feature coding vector of the indoor occupancy state to obtain an optimized sequence of the target offset-dynamic gain local association response coding matrices; inputting the optimized sequence of the target offset-dynamic gain local association response coding matrices into an LSTM transfer coding module with an attention mechanism to obtain the target offset-dynamic gain fine-grained collaborative coding vector.

[0052] Particularly, considering that the segmentation granularity of the feature segmentation function affects the local region size of the target offset-dynamic gain local association response coding matrix, it will also directly affect the expression of the feature interaction pattern between the corresponding local embedding feature coding vector of the carbon dioxide concentration target offset amount and the local embedding feature coding vector of the indoor occupancy state. That is to say, since the target offset-dynamic gain local association response coding matrix is a transition response between two local regions, it essentially expresses the spatial measure of the transition response space based on its row vectors, and the row vector length is also the representation of the local region size of the above-mentioned target offset-dynamic gain local association response coding matrix. If the local region size of the target offset-dynamic gain local association response coding matrix, that is, the row vector length L, is introduced as the local region size intensity constraint, then the sparse spatial measure expression of the target offset-dynamic gain local association response coding matrix, that is, the F-norm ||M i || F should follow a Poisson-like relationship:

[0053]

[0054] Among them, (!) represents factorial, and ||·|| F represents the Frobenius norm, L represents the field size of M i , e (·) represents the exponential function with the natural constant as the base, and λ represents the sparse space measure expression parameter. That is, the row vector length L of the target offset-dynamic gain local correlation response coding matrix acts as the local size intensity constraint on the sparse space measure of the local region transfer response space for L times.

[0055] Thus, the parameter λ can be solved. In this way, when the transfer response interaction is generated by introducing a Poisson-like process with intensity L and mean expectation λ, using ||x i ||2 and ||y i ||2 to describe the edge connection representation of the space measure, the transfer response connection probability between two local regions is further determined as:

[0056]

[0057] where ρ represents the transfer response connection probability density between x i and y i , |·| represents calculating the modulus length, and ||·||2 represents calculating the second norm.

[0058] Then, the local region size L is iteratively corrected using the transfer response connection probability:

[0059] L′ = ρL

[0060] where L′ represents the asymptotically optimized L.

[0061] That is, while strictly ensuring that the expected degree is L, the regularization of the overall space measure is determined by the regularity constraint of the mean expectation connection probability fluctuation of each row, so that the structured interaction information within the local space measure can avoid local overfitting. Finally, the sparsity of the target offset-dynamic gain local correlation response coding matrix M i is re-constrained with the optimized L′:

[0062]

[0063] where M′ i represents the optimized target offset-dynamic gain local correlation response coding matrix corresponding to M i .

[0064] That is, by forcing the sparse representation of the target offset-dynamic gain local correlation response encoding matrix to match the local size L′, the feature intensity and sparsity are balanced, and the overall expression effect of the sequence of the target offset-dynamic gain local correlation response encoding matrix is improved.

[0065] Specifically, in step S324, the target offset-dynamic gain fine-grained collaborative coding vector is subjected to feature decoding to obtain controller optimization parameters, where the controller optimization parameters include an optimization ratio parameter, an optimization integral parameter, and an optimization differential parameter. In a specific example of the present application, step S324 includes: inputting the target offset-dynamic gain fine-grained collaborative coding vector into a feature decoder based on a multi-layer perceptron to obtain the controller optimization parameters. It should be understood that the target offset-dynamic gain fine-grained collaborative coding vector reveals the deep collaborative relationship between the current carbon dioxide concentration offset and the indoor occupancy state characteristics, and implies the optimal dynamic response characteristics that the control system should possess under the combination of the current carbon dioxide concentration offset and the indoor occupancy state. Therefore, in order to further quantify the optimal dynamic response characteristics implied by the target offset-dynamic gain fine-grained collaborative coding vector and convert it into controller parameters that can be directly applied to the fresh air control system, the present application constructs a feature decoder based on a multi-layer perceptron to perform feature decoding processing on the target offset-dynamic gain fine-grained collaborative coding vector. Through multi-layer non-linear transformation, the high-dimensional collaborative coding vector is gradually mapped into a low-dimensional controller parameter space, thereby achieving an accurate solution for the controller optimization parameters. During the decoding process, each layer of the perceptron will perform weighted summation and non-linear activation operations according to the input features of the current layer to extract deeper feature information and gradually approach the optimal controller parameters. Finally, at the output layer of the decoder, through the regression operation of the neural network, the specific values of the optimization ratio parameter, the optimization integral parameter, and the optimization differential parameter are obtained, guiding the dynamic adjustment of the fresh air control system to achieve more accurate and efficient indoor air quality control.

[0066] Specifically, in step S33, the controller optimization parameters are updated to the parameter register of the fresh air volume PID controller to obtain an optimized fresh air volume PID controller. Specifically, in order to construct an online adaptive control architecture, the present application is based on the dynamic reconfiguration principle of a field programmable gate array (FPGA), and realizes real-time parameter refreshing through register mapping technology. Specifically, a dedicated parameter storage area is opened in the embedded controller, and the generated controller optimization parameters are written to the corresponding register address through the Modbus communication protocol; the controller reads the latest parameters at the beginning of each control cycle (such as 1 second) and applies them to the current control law calculation. In this way, parameter iterative update can be completed on a second-level time scale, ensuring that the controller always matches the latest environmental state.

[0067] Specifically, in step S34, the carbon dioxide concentration target offset is input into the optimized fresh air volume PID controller to obtain the basic fresh air volume adjustment command. It should be understood that since the dynamically optimized PID controller needs to generate a control quantity in combination with the real-time deviation signal. Therefore, in order to complete the fresh air volume adjustment in a closed loop, the present application is based on the incremental PID algorithm principle, and calculates the fresh air mass flow value that needs to be introduced through the optimized control law. Specifically, the carbon dioxide concentration target offset is input into the optimized fresh air volume PID controller. According to the incremental PID control algorithm, the controller calculates the fresh air mass flow value that should be introduced based on the deviation between the current carbon dioxide concentration and the target concentration, in combination with the optimized parameters of the proportional, integral, and differential links, so as to ensure that the fresh air control system can perform fast and accurate adjustment according to the real-time environmental state, maintain the indoor air quality within the preset range, and provide a comfortable and healthy environment for the indoor personnel.

[0068] In the above-mentioned linkage control method for energy-saving ventilation equipment, in step S4, based on the deviation between the real-time indoor temperature value and the temperature target value and the indoor occupancy state data, the basic air-conditioning load adjustment command is calculated. It should be understood that since maintaining indoor thermal comfort is the basic responsibility of the air-conditioning system, it is necessary to adjust the cooling or heating output according to the difference between the actual indoor temperature and the desired temperature; at the same time, considering the energy-saving requirements, the operating intensity and target temperature of the air-conditioning system should be adjusted according to whether there are people in the room to avoid maintaining strict comfort temperature standards when there is no one. Therefore, in order to generate a preliminary air-conditioning adjustment command that reflects the basic heat load demand according to the current indoor thermal environment conditions and occupancy state, as the basis for the temperature control loop in the subsequent linkage control. The present application is based on the classical temperature feedback control principle and combines an energy-saving control strategy based on the occupancy state (such as temperature setpoint floating) to calculate the basic air-conditioning load adjustment command. Specifically, first, the indoor occupancy state is judged. If there is no one in the room, it is automatically switched to the energy-saving standby temperature target value (for example, in summer, the temperature is allowed to rise to 28 - 30 °C, and in winter, the temperature is allowed to drop to 16 - 18 °C, forming a relatively wide temperature dead zone); if there are people in the room, the comfortable temperature target value for the personnel is used (for example, 24 - 26 °C in summer and 20 - 22 °C in winter). Then, the deviation (i.e., temperature error) between the real-time indoor temperature value and the current effective temperature target value is calculated, and the obtained temperature error is input into the temperature PID controller. According to the magnitude of the temperature error, the total heat (cooling load) that needs to be removed or the heat (heating load) that needs to be supplemented in the current space is calculated to obtain the basic air-conditioning load adjustment command. In this way, it is ensured that a comfortable temperature is provided when there are people, and the energy-saving mode is automatically entered when there is no one, constituting the core feedback part of the temperature control loop.

[0069] In the above-mentioned linkage control method for energy-saving ventilation equipment, in step S5, based on the basic fresh air volume adjustment instruction and the difference between the outdoor temperature value and the real-time indoor temperature value, the disturbance load introduced by the fresh air volume is estimated. It should be understood that since the outdoor fresh air introduced by the ventilation system (whose temperature is usually different from that of the indoor) will inevitably have a direct impact on the indoor thermal environment, forming an additional cooling or heating load, which is an important reason for the lag of the air-conditioning system response and the increase in energy consumption in the traditional control method. Therefore, in order to quantify this upcoming thermal disturbance and provide a basis for subsequent feedforward compensation in air-conditioning control, thereby improving the stability of temperature control and the overall energy efficiency of the system. This application is based on the basic principles of thermodynamics, that is, the heat transfer is calculated through the mass flow rate of air, specific heat capacity, and the temperature difference between indoor and outdoor. Specifically, based on the basic fresh air volume adjustment instruction (the intended fresh air mass flow rate value) obtained from the above calculation and the difference between the outdoor temperature value and the real-time indoor temperature value (ΔT = T_outdoor - T_indoor), the heat load calculation formula is applied: where the specific heat capacity of air can be regarded as a constant. Finally, the estimated value of the disturbance load introduced by the fresh air volume is obtained. This value reflects the additional load amount on the indoor thermal environment due to the introduction of fresh air. When it is a positive number, it means that the fresh air brings a heating load (it is necessary to increase cooling or reduce heating); when it is a negative number, it means that the fresh air brings a cooling load (it is necessary to increase heating or reduce cooling).

[0070] In the above-mentioned linkage control method for energy-saving ventilation equipment, in step S6, the disturbance load introduced by the fresh air volume is input as a feedforward signal into the temperature control loop to obtain the final air-conditioning load adjustment instruction. It should be understood that due to the lag of traditional feedback control, it is difficult to immediately offset the sudden load disturbance introduced by the fresh air. Therefore, in order to construct a feedforward-feedback composite control architecture, this application is based on the dynamic feedforward linearization technology, and the anti-interference ability of the temperature control loop is improved through the pre-compensation mechanism of the disturbance load. Specifically, the disturbance load value introduced by the fresh air volume is algebraically superimposed with the basic air-conditioning load adjustment instruction, and the output power of the air-conditioning system is directly corrected through the feedforward channel, thereby obtaining the final air-conditioning load adjustment instruction, which helps to respond to the load disturbance introduced by the fresh air in advance, effectively reduce the indoor temperature fluctuation, improve the stability and response speed of the control system, significantly reduce the overshoot and oscillation phenomena, and realize the dynamic coordination of the ventilation and air-conditioning systems.

[0071] In summary, the linkage control method for energy-saving ventilation equipment according to the embodiments of the present application is clarified. It collects data on indoor and outdoor temperatures, indoor carbon dioxide concentration, and occupancy status in real time, and through in-depth collaborative analysis of the deviation between the indoor occupancy status, carbon dioxide concentration, and the target set value, it realizes the joint characterization of the current indoor fresh air volume regulation demand, and accordingly dynamically optimizes the control parameters of the fresh air volume PID controller to calculate the basic fresh air volume adjustment instruction. On this basis, based on the temperature difference between the outdoor and indoor, the heat disturbance introduced by fresh air is further quantified and used as a feedforward signal to compensate the air-conditioning load regulation loop to achieve dynamic adjustment of the air-conditioning load. In this way, the coordinated control of the ventilation and air-conditioning systems can be realized, suppressing the energy consumption coupling conflict between the fresh air and air-conditioning equipment in traditional control, so as to reduce building energy consumption while ensuring the indoor environmental quality.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A linkage control method for energy-saving ventilation equipment, characterized in that, Including: Obtain the real-time indoor carbon dioxide concentration value and the real-time indoor temperature value; Obtain the outdoor temperature value and the indoor occupancy status data; Calculate a basic fresh air volume adjustment instruction based on the deviation between the real-time indoor carbon dioxide concentration value and the carbon dioxide target value and the indoor occupancy status data; Calculate a basic air-conditioning load adjustment instruction based on the deviation between the real-time indoor temperature value and the temperature target value and the indoor occupancy status data; Estimate the disturbance load introduced by the fresh air volume based on the basic fresh air volume adjustment instruction and the difference between the outdoor temperature value and the real-time indoor temperature value; Input the disturbance load introduced by the fresh air volume as a feedforward signal into the temperature control loop to obtain the final air-conditioning load adjustment instruction.

2. The linkage control method for the energy-saving ventilation equipment according to claim 1, characterized in that, Calculating a basic fresh air volume adjustment instruction based on the deviation between the real-time indoor carbon dioxide concentration value and the carbon dioxide target value and the indoor occupancy status data includes: Calculate the deviation between the real-time indoor carbon dioxide concentration value and the carbon dioxide target value to obtain a carbon dioxide concentration target offset; Adaptive optimize the fresh air volume control parameters based on the deep collaborative association response characteristics between the carbon dioxide concentration target offset and the indoor occupancy status data to obtain controller optimization parameters; Update the controller optimization parameters to the parameter register of the fresh air volume PID controller to obtain an optimized fresh air volume PID controller; Input the carbon dioxide concentration target offset into the optimized fresh air volume PID controller to obtain the basic fresh air volume adjustment instruction.

3. The linkage control method for an energy-saving ventilation device according to claim 2, characterized in that Adaptive optimize the fresh air volume control parameters based on the deep collaborative association response characteristics between the carbon dioxide concentration target offset and the indoor occupancy status data to obtain controller optimization parameters, including: Perform one-hot encoding on the carbon dioxide concentration target offset to obtain a one-hot embedded encoding vector of the carbon dioxide concentration target offset; Perform structured embedding encoding on the indoor occupancy status data to obtain a structured embedded encoding vector of the indoor occupancy status; Perform fine-grained collaborative response association analysis on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy status to obtain a target offset-dynamic gain fine-grained collaborative encoding vector; Perform feature decoding on the target offset-dynamic gain fine-grained collaborative encoding vector to obtain controller optimization parameters, where the controller optimization parameters include an optimized proportional parameter, an optimized integral parameter, and an optimized derivative parameter.

4. The linkage control method for an energy-saving ventilation device according to claim 3, wherein Performing fine-grained collaborative response association analysis on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy status to obtain a target offset-dynamic gain fine-grained collaborative encoding vector includes: Perform feature association response inference based on local granularity on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy status to obtain a sequence of target offset-dynamic gain local association response encoding matrices; Perform sequence transfer encoding on the sequence of target offset-dynamic gain local association response encoding matrices to obtain the target offset-dynamic gain fine-grained collaborative encoding vector.

5. The linkage control method for the energy-saving ventilation equipment according to claim 4, characterized in that, Performing feature association response inference based on local granularity on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy state to obtain a sequence of target offset-dynamic gain local association response encoding matrices, including: Performing equal-granularity feature segmentation on the one-hot embedded encoding vector of the carbon dioxide concentration target offset and the structured embedded encoding vector of the indoor occupancy state to obtain a sequence of local embedded feature encoding vectors of the carbon dioxide concentration target offset and a sequence of local embedded feature encoding vectors of the indoor occupancy state; Inputting each pair of corresponding local embedded feature encoding vectors of the carbon dioxide concentration target offset and the local embedded feature encoding vectors of the indoor occupancy state in the sequences of the local embedded feature encoding vectors of the carbon dioxide concentration target offset and the local embedded feature encoding vectors of the indoor occupancy state into a transfer response inference unit to obtain the sequence of target offset-dynamic gain local association response encoding matrices.

6. The linkage control method for an energy-saving ventilation device according to claim 5, characterized in that, Performing sequence transfer encoding on the sequence of target offset-dynamic gain local association response encoding matrices to obtain the target offset-dynamic gain fine-grained collaborative encoding vector, including: Inputting the sequence of target offset-dynamic gain local association response encoding matrices into an LSTM transfer encoding module with an attention mechanism to obtain the target offset-dynamic gain fine-grained collaborative encoding vector.

7. The linkage control method for an energy-saving ventilation device according to claim 6, characterized in that, Inputting the sequence of target offset-dynamic gain local association response encoding matrices into an LSTM transfer encoding module with an attention mechanism to obtain the target offset-dynamic gain fine-grained collaborative encoding vector, including: Performing flattening processing on each target offset-dynamic gain local association response encoding matrix in the sequence of target offset-dynamic gain local association response encoding matrices to obtain a sequence of target offset-dynamic gain local association response encoding vectors; Using the Frobenius norm of each target offset-dynamic gain local association response encoding matrix in the sequence of target offset-dynamic gain local association response encoding matrices as a constraint factor to perform feature modulation on the sequence of target offset-dynamic gain local association response encoding vectors to obtain a sequence of modulated target offset-dynamic gain local association response encoding vectors; Performing transfer encoding based on the LSTM model on the sequence of modulated target offset-dynamic gain local association response encoding vectors to obtain the target offset-dynamic gain fine-grained collaborative encoding vector.

8. The linkage control method for the energy-saving ventilation equipment according to claim 5, characterized in that, Performing sequence transfer encoding on the sequence of target offset-dynamic gain local association response encoding matrices to obtain the target offset-dynamic gain fine-grained collaborative encoding vector, including: Performing sparse regularization constraint on each target offset-dynamic gain local association response encoding matrix in the sequence of target offset-dynamic gain local association response encoding matrices based on the transfer response connection probability between each pair of corresponding local embedded feature encoding vectors of the carbon dioxide concentration target offset and the local embedded feature encoding vectors of the indoor occupancy state to obtain a sequence of optimized target offset-dynamic gain local association response encoding matrices; Input the sequence of the optimized target offset-dynamic gain local correlation response encoding matrix into the LSTM transfer encoding module with an attention mechanism to obtain the target offset-dynamic gain fine-grained collaborative encoding vector.

9. The linkage control method for an energy-saving ventilation device according to claim 3, characterized in that Perform feature decoding on the target offset-dynamic gain fine-grained collaborative encoding vector to obtain the controller optimization parameters, including: Input the target offset-dynamic gain fine-grained collaborative encoding vector into the feature decoder based on a multi-layer perceptron to obtain the controller optimization parameters.

Images