Air quality control method, system and equipment for underground space environment and medium

Abstract

The invention discloses an air quality control method, system and device for an underground space environment and a medium, and relates to the technical field of air quality control. The method comprises the steps that all influence indexes influencing the air quality in a monitoring area are obtained through a preset sensor, obtaining a target quality grade of the monitoring area through a preset algorithm model or a network model based on each influence index; controlling the fresh air ventilator to operate in the corresponding fresh air ventilator working state according to the target quality grade on the basis of the preset fresh air ventilator working states of different grades; after the target quality grade of the monitoring area is obtained, the fresh air ventilator working states are formulated based on the different air quality grades, and the fresh air ventilator is controlled to operate in the fresh air ventilator working states of the corresponding grades; on one hand, the dependence on experience requirements of workers is solved, and on the other hand, energy conservation and ventilation efficiency are greatly improved by accurately determining the air quality grade and operating in the corresponding fresh air ventilator working state.

Description

Technical Field

[0001] The present invention relates to the technical field of air quality control, and more specifically, to an air quality control method, system, equipment and medium for underground space environment. Background Art

[0002] The underground caverns of underground hydropower stations are intricate, the underground plant is large, and the structure is irregular. The building structure is complex, and the local airflow organization is difficult to accurately control (such as the generator installation room and the terminal, busbar hole, main transformer room, etc.). In summer, the main transformer room and busbar room have poor air circulation, and the indoor temperature is as high as 38°C, which seriously affects the safe operation of indoor equipment. In addition, the outdoor air state parameters and the wet load of the power station enclosure structure vary greatly. Condensation occurs in some caverns, causing mechanical equipment to rust, electrical equipment to reduce insulation capacity, leakage, breakdown and other problems, which seriously affect the safe operation of the equipment; especially for air quality, it will seriously affect the personal safety of the staff, so a device that can detect and control air quality is needed.

[0003] At present, the air quality of the space environment is usually detected by various sensors and fed back to the staff. The staff adjusts the operating power of the fan according to the detection results to replace the air in the space environment to improve the air quality. On the one hand, the staff's judgment on the detection results requires a high level of experience. On the other hand, the judgment results of the detection results will directly affect the power consumption and working efficiency of the fan. If the judgment is inaccurate, the power consumption will increase or the air replacement efficiency will be low. At the same time, there is also the use of network models to judge the air quality, but there are also problems such as long model training time and low accuracy.

[0004] In view of this, this application is hereby made. Summary of the invention

[0005] The purpose of the present invention is to provide an air quality control method, system, equipment and medium for an underground space environment to solve the problems existing in the above-mentioned background technology.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions:

[0007] In a first aspect, the present application provides an air quality control method for an underground space environment, comprising the following specific steps:

[0008] Obtain various influencing indicators that affect the air quality in the monitoring area through preset sensors, and obtain the target quality level of the monitoring area based on various influencing indicators and through preset algorithm models or network models;

[0009] Based on the preset working states of the fresh air fans at different levels, the fresh air fans are controlled to operate in the corresponding working states according to the target quality level; wherein the target quality level of the monitoring area is obtained through the algorithm model, specifically including:

[0010] Based on the number of preset different levels and various influencing indicators, the degree of membership of each influencing indicator to each level and the influence weight of each influencing indicator are calculated;

[0011] According to the membership degree and influence weight of each influencing indicator to each level, the target quality level of the monitoring area is calculated based on the preset fuzzy operator;

[0012] The target quality level of the monitoring area is obtained through the network model, including:

[0013] The preset initial model is optimized and trained using the historical indicator data of the monitoring area and the historical levels corresponding to the historical indicator data, so as to obtain a network model that has been optimized and trained and corresponds to the preset initial model;

[0014] Each influencing indicator is input into the network model for processing, and the network model outputs the target quality level corresponding to each influencing indicator.

[0015] Based on the above technical solution, the present invention can also be improved as follows.

[0016] Furthermore, the above optimization training includes: introducing a momentum factor in the iterative training of the preset initial model to adjust the model weight of the preset initial model, and the model parameters include the model weight, specifically:

[0017]

[0018] Where Δw(k+1) represents the change in model weight at the k+1th iteration, η represents the momentum factor, 0<η<1, Δw(k) represents the change in model weight at the kth iteration, α represents the model learning rate of the current iteration, E(k) represents the value of the error function of the preset initial model at the kth iteration, E(k)=1 / 2Σ(A(k)-B(k)) 2 , A(k) represents the actual output value of the preset initial model at the kth iteration, and B(k) represents the expected output value of the preset initial model at the kth iteration. Represents the partial derivative of the error function with respect to the model weights.

[0019] Furthermore, the above model parameters include a model learning rate, which is specifically:

[0020]

[0021] Wherein, α(k+1) represents the model learning rate at the k+1th iteration, α(k) represents the model learning rate at the kth iteration, E(k) represents the value of the error function of the preset initial model at the kth iteration, and E(k+1) represents the value of the error function of the preset initial model at the k+1th iteration.

[0022] Furthermore, the above model weights are modified in the following way:

[0023]

[0024] In the formula, Δw(k+1) represents the change in the model weight at the k+1th iteration, Δw(k) represents the change in the model weight at the kth iteration, σ represents the increment factor, and σ>1, β represents the decrement factor, and β<1, Represents the partial derivative of the error function with respect to the model weights.

[0025] Furthermore, the above model weights are modified in the following way:

[0026] Δw(k+1)=Δw(k)-(J T +μI) -1 J T e;

[0027] Where Δw(k+1) represents the change in the model weight at the k+1th iteration, 0<η<1, Δw(k) represents the change in the model weight at the kth iteration, J T Represents the Jacobian matrix, where the elements are the partial derivatives of the error function with respect to the weights, μ represents the correlation coefficient, I represents the identity matrix, and e represents the error vector.

[0028] Furthermore, the degree of membership of each of the above-mentioned influencing indicators to each level is as follows:

[0029] When j = 1,

[0030] When j = 2, 3, 4...n-1,

[0031] When j = n,

[0032] In the formula, j represents the serial number of each different level, j = 1, 2, 3...n, r ij represents the degree of membership of the i-th influencing indicator to the j-th level, x i represents the measured value of the i-th impact indicator, S ij It represents the standard value of the i-th impact indicator corresponding to level j.

[0033] Furthermore, the impact weight of each of the above impact indicators is as follows:

[0034]

[0035] In the formula, a i represents the influence weight of the i-th influencing indicator, x i represents the measured value of the i-th impact indicator, S i It represents the mean of the standard values ​​of the i-th influencing indicator in each level.

[0036] In a second aspect, the present application provides an air quality control system for an underground space environment, which is applied to an air quality control method for an underground space environment according to any one of the first aspects, comprising:

[0037] The first module is used to obtain various influencing indicators that affect the air quality in the monitoring area through preset sensors, and obtain the target quality level of the monitoring area based on the various influencing indicators and through a preset algorithm model or network model;

[0038] The second module is used to control the fresh air fan to operate in the corresponding fresh air fan working state based on the preset working state of each different level and according to the target quality level; wherein the target quality level of the monitoring area is obtained by the algorithm model in the first module, specifically including:

[0039] The first submodule is used to calculate the degree of membership of each influencing indicator to each level and the influence weight of each influencing indicator based on the number of preset different levels and each influencing indicator;

[0040] The second submodule is used to calculate the target quality level of the monitoring area based on the preset fuzzy operator according to the membership degree and influence weight of each influencing indicator to each level;

[0041] In the first module, the target quality level of the monitoring area is obtained through the network model, including:

[0042] The third submodule is used to optimize and train the preset initial model using the historical indicator data of the monitoring area and the historical levels corresponding to the historical indicator data, so as to obtain a network model that has been optimized and trained and corresponds to the preset initial model;

[0043] The fourth submodule is used to input various influencing indicators into the network model for processing, and the network model outputs the target quality level corresponding to each influencing indicator.

[0044] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any one of the methods in the first aspect when executing the computer program.

[0045] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in the first aspect.

[0046] Compared with the prior art, the present invention has at least the following beneficial effects:

[0047] In the present application, firstly, various influencing indicators affecting the air quality are obtained in the monitoring area through preset sensors. Each influencing indicator can be determined in advance, such as carbon dioxide, PN2.5, carbon monoxide, etc., and each different level can also be determined in advance, and it is necessary to clarify the number of levels, the numerical range of each influencing indicator in each level, etc.; secondly, the air quality level of the monitoring area is obtained through an algorithm model or a network model, that is, the target quality level of the monitoring area; finally, based on the working state of the fresh air fan formulated for each different air quality level, the fresh air fan is controlled to operate in the fresh air fan working state of the corresponding level; on the one hand, this solves the dependence on the experience requirements of the staff, and on the other hand, by accurately determining the air quality level and operating it in the corresponding fresh air fan working state, the energy saving and ventilation efficiency are also greatly improved.

[0048] In the present application, when the air quality level in the monitoring area is predicted by the algorithm model, the degree of membership of each influencing indicator to each level and the influence weight of each influencing indicator are used to calculate the final corresponding air quality level, which is easy to implement in actual implementation and has high prediction accuracy; and when the air quality level in the monitoring area is predicted by the network model, each influencing indicator is input into the network model for processing, and the network model outputs the target quality level corresponding to each influencing indicator, which is convenient and fast, and the network model is obtained through optimization training. During the optimization training process of the network model, the model weights of the preset initial model are optimized and corrected by introducing momentum factors, incremental factors, decrement factors, etc., and the model learning rate of the preset initial model is also optimized and corrected. In this way, on the one hand, the final network model can be more accurate in predicting the air quality level, and on the other hand, the preset initial model can converge faster during the training process and learn the relationship between features more efficiently, thereby completing the model training efficiently and quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0050] Figure 1 A method flow chart of a control method in an embodiment of the present invention;

[0051] Figure 2 This is a flow chart of obtaining the target quality level of the monitoring area through the algorithm model in an embodiment of the present invention;

[0052] Figure 3 A flow chart of obtaining a target quality level of a monitoring area through a network model in an embodiment of the present invention;

[0053] Figure 4 This is a connection diagram of a control system in an embodiment of the present invention;

[0054] Figure 5 Schematic diagram of the connection of electronic equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0056] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0058] In the description of the embodiments of the present invention, "plurality" means at least 2.

[0059] Embodiment 1:

[0060] This embodiment provides an air quality control method for an underground space environment, such as Figure 1 As shown, the following specific steps are included:

[0061] S1, obtaining various influencing indicators affecting the air quality in the monitoring area through preset sensors, and obtaining the target quality level of the monitoring area based on the various influencing indicators and through a preset algorithm model or network model.

[0062] Among them, various influencing indicators can be determined in advance, for example, they can be carbon dioxide, PN2.5, carbon monoxide, etc., and the prefabricated sensors can be carbon dioxide concentration sensors, PN2.5 concentration sensors, carbon monoxide concentration sensors, etc.; and the different levels can also be determined in advance, and it is necessary to clarify the number of levels, the numerical range of each influencing indicator in each level, etc. For example, the levels can be divided into four levels: unpolluted, lightly polluted, moderately polluted and heavily polluted; and correspondingly, the number of working states of the fresh air fan corresponds to the number of divided levels, that is, it is also four.

[0063] S2, based on the preset working states of the fresh air fans at different levels, and according to the target quality level, the fresh air fans are controlled to operate in the corresponding working states of the fresh air fans.

[0064] Among them, the working state of the fresh air fan is formulated for each different air quality level, so as to control the fresh air fan to operate in the corresponding level of fresh air fan working state; on the one hand, this solves the dependence on the experience requirements of the staff, and on the other hand, by accurately determining the air quality level and operating it in the corresponding fresh air fan working state, the energy saving and ventilation efficiency are also greatly improved.

[0065] It should be noted that other state parameters in the monitoring area can also be controlled, such as temperature and humidity, that is, by installing corresponding temperature and humidity sensors in the monitoring area, after detecting the temperature and humidity data, the humidifier, dryer, air cooler, etc. are controlled to operate, thereby improving the temperature and humidity state in the monitoring area. Specifically, through the above scheme: 1. The ambient temperature of the plant can be reduced, and the ventilation volume of the plant can be increased, thereby reducing the operating temperature of the main transformer, knife switch, busbar and other equipment by more than 5°C. 2. Improve the dust problem in the underground plant, improve the air quality, reduce equipment failures such as short circuit and discharge caused by excessive dust, reduce the amount of dust inhaled by operators, and improve the working environment. 3. The room temperature of the main equipment such as the main transformer tunnel, busbar tunnel, ventilation duct, etc. is controlled below 30°C all year round, and the humidity does not exceed 85%. This scheme has the characteristics of low cost, complete functions, high automation and intelligence, can meet the production and operation requirements of the hydropower station, and is conducive to the realization of the production management mode of intelligent air quality monitoring and control of the hydropower station.

[0066] Among them, when adjusting the temperature and humidity, the following plan can be adopted, which includes: analyzing system requirements according to actual on-site conditions; building control systems and transmitters in hardware equipment; building lower-level computer programs, developing upper-level computer software, and then jointly debugging with the hardware equipment; jointly debugging the entire system, writing a construction plan for the online monitoring system, and conducting on-site construction and debugging.

[0067] Among them: the hardware equipment adopts a distributed architecture, which is divided into three layers, including the management layer, the middle layer and the field layer; the management layer includes data transmission and fan control, the upper computer software system running in the industrial computer for online monitoring, the configuration software as a development tool, and the design of the human-computer interaction interface to achieve real-time monitoring, remote control and fault alarm. The middle layer uses PLC to start and stop each fan, and transmits the real-time data collected by different environmental parameter transmitters to the industrial computer of the management layer. At the same time, the master station PLC is equipped with a touch screen, which reads the real-time data of environmental parameters and realizes remote control of the fan through the touch screen. Ethernet or optical fiber is used to connect the PLCs. Each PLC is responsible for a designated fixed area. Ethernet is used to connect the middle layer and the management layer to form a distributed online monitoring system. The field layer is the bottom layer, which includes the layout of various transmitters, fans and relays; the transmitters can be divided into temperature and humidity type, wind speed type, and oxygen type; the sensors used in the online test are all industrial-grade transmitters, and the sensors are devices or devices that can convert the specified measured quantity into a usable output signal according to a certain rule.

[0068] The intelligent transformation of the ventilation system of the underground powerhouse of the hydropower station includes: analyzing the control strategy of the supply / induced draft fan, determining the object of environmental monitoring; selecting PLC and transmitter respectively; writing and debugging the operation program. If the operation requirements are met, the online monitoring system is combined for cascade debugging; if the operation requirements are not met, the operation program is modified and debugged again until the cascade debugging requirements are met. Build the wiring diagram and control cabinet of the automatic control system; install the automatic control system and transmitter; perform cascade debugging with the operation program. If the requirements are met, write the transformation plan of the ventilation system, and perform on-site installation and debugging; if the requirements are not met, perform hardware debugging until the requirements are met and the transformation plan of the ventilation system can be written.

[0069] Furthermore, the above algorithm model is used to obtain the target quality level of the monitoring area, such as Figure 2 As shown, specifically including:

[0070] S111, based on the preset number of different levels and each influencing indicator, the degree of membership of each influencing indicator to each level and the influence weight of each influencing indicator are calculated.

[0071] Optionally, the degree of membership of each of the above-mentioned influencing indicators to each level is as follows:

[0072] When j = 1,

[0073] When j = 2, 3, 4...n-1,

[0074] When j = n,

[0075] In the formula, j represents the serial number of each different level, j = 1, 2, 3...n, r ij represents the degree of membership of the i-th influencing indicator to the j-th level, x i represents the measured value of the i-th impact indicator, S ij It represents the standard value of the i-th impact indicator corresponding to level j.

[0076] Wherein, j represents the serial number of each different level, and j=1, 2, 3...n. In the above calculation formula, the calculation of the membership degree of each influencing indicator corresponding to the first and last two levels is different from other calculations.

[0077] Optionally, the influence weight of each of the above-mentioned influence indicators may be:

[0078]

[0079] In the formula, a i represents the influence weight of the i-th influencing indicator, x i represents the measured value of the i-th impact indicator, S i It represents the mean of the standard values ​​of the i-th influencing indicator in each level.

[0080] S112, according to the membership degree and influence weight of each influencing indicator to each level, the target quality level of the monitoring area is calculated based on a preset fuzzy operator.

[0081] The fuzzy operator can be After fuzzy operator calculation, several groups of values ​​corresponding to the number of levels can be obtained. The largest of these values ​​is the final coefficient of the monitoring area. Then see which level the value corresponds to. The corresponding level is the final predicted level of the detection area, that is, the target quality level. For example, after the degree of membership and influence weight of each level by each influencing indicator and the fuzzy operator calculation, there are four levels of "unpolluted, lightly polluted, moderately polluted and heavily polluted" as mentioned above, and the calculated values ​​can be "45.55, 63.22, 52.36, 46.88", among which 45.55 corresponds to unpolluted, 63.22 corresponds to lightly polluted, 52.36 corresponds to moderately polluted, and 46.88 corresponds to heavily polluted, that is, the order of the obtained values ​​is the same as the order of the levels. Among these values, the largest is 63.22, and the level corresponding to this value is lightly polluted, which indicates that the final predicted level of the monitoring area is lightly polluted, that is, the target quality level is lightly polluted.

[0082] Furthermore, the target quality level of the monitoring area is obtained through the network model, such as Figure 3 As shown, specifically including:

[0083] S121, optimizing and training the preset initial model using the historical indicator data of the monitoring area and the historical levels corresponding to the historical indicator data, to obtain a network model that has been optimized and trained and corresponds to the preset initial model.

[0084] Among them, the above-mentioned network model can make a bp neural network model. The steps during optimization training are: input historical indicator data into the model, the model will output the prediction results corresponding to the historical indicator data, and then use the prediction results and the historical levels corresponding to the historical indicator data to calculate the model loss function, and continuously adjust the model parameters of the model according to the size of the loss function until the training end conditions are met and the model training is completed.

[0085] Optionally, the above optimization training includes: introducing a momentum factor in the iterative training of the preset initial model to adjust the model weight of the preset initial model, and the model parameters include the model weight, which may be specifically:

[0086] in,

[0087] Where Δw(k+1) represents the change in model weight at the k+1th iteration, η represents the momentum factor, 0<η<1, Δw(k) represents the change in model weight at the kth iteration, α represents the model learning rate of the current iteration, E(k) represents the value of the error function of the preset initial model at the kth iteration, E(k)=1 / 2Σ(A(k)-B(k)) 2 , A(k) represents the actual output value of the preset initial model at the kth iteration, and B(k) represents the expected output value of the preset initial model at the kth iteration. Represents the partial derivative of the error function with respect to the model weights.

[0088] Optionally, the above model parameters include a model learning rate, which may be:

[0089]

[0090] Wherein, α(k+1) represents the model learning rate at the k+1th iteration, α(k) represents the model learning rate at the kth iteration, E(k) represents the value of the error function of the preset initial model at the kth iteration, and E(k+1) represents the value of the error function of the preset initial model at the k+1th iteration.

[0091] Optionally, the above model weights can also be modified in the following ways:

[0092]

[0093] In the formula, Δw(k+1) represents the change in the model weight at the k+1th iteration, Δw(k) represents the change in the model weight at the kth iteration, σ represents the increment factor, and σ>1, β represents the decrement factor, and β<1, Represents the partial derivative of the error function with respect to the model weights.

[0094] Optionally, the above model weights can also be modified in the following ways:

[0095] Δw(k+1)=Δw(k)-(J T +μI) -1 J T e;

[0096] Where Δw(k+1) represents the change in the model weight at the k+1th iteration, 0<η<1, Δw(k) represents the change in the model weight at the kth iteration, J T Represents the Jacobian matrix, where the elements are the partial derivatives of the error function with respect to the weights, μ represents the correlation coefficient, I represents the identity matrix, and e represents the error vector.

[0097] Among them, the following problems exist in the training process of the network model: 1. There is no definite derivation theory for the number of layers and nodes of the hidden layer, and a large number of experiments are required to obtain the ideal number of layers and nodes of the hidden layer. Therefore, the structural design of the network is sometimes not the most appropriate. If the number of layers and nodes of the hidden layer is too large, the network learning efficiency will be reduced, the convergence time of the network will be increased, and even overfitting will occur; if the number of layers and nodes of the hidden layer is too small, the network will not converge, and then the indoor air quality will be wrongly evaluated; 2. The selection of the learning rate also requires a lot of experimental experience to obtain. If the selected learning rate is too large, the network will not converge and the expected error will not be achieved; if the selected learning rate is too small, the number of training steps of the network will increase, resulting in an increase in the convergence time of the network. Therefore, it is difficult to select the best learning rate suitable for the network; 3. Since the traditional BP neural network is based on the fastest gradient descent method, this method converges slowly and is "greedy", and it is easy to fall into local extreme values ​​and thus fail to obtain the optimal results, causing the network to make incorrect evaluations of indoor air quality; 4. The network training process is unstable. When appropriate parameters are selected, the error accuracy required by the network can be achieved, but it cannot guarantee that the target error accuracy can be achieved in each training; these problems can be solved by adjusting the model parameters as mentioned above.

[0098] S122, input each influencing indicator into the network model for processing, and the network model outputs the target quality level corresponding to each influencing indicator.

[0099] Among them, when the air quality level in the monitoring area is predicted by the network model, each influencing index is input into the network model for processing, and the network model outputs the target quality level corresponding to each influencing index, which is convenient and fast.

[0100] Embodiment 2:

[0101] This embodiment provides an air quality control system for an underground space environment, which is applied to any of the air quality control methods for an underground space environment in Embodiment 1, such as Figure 4 As shown, it may include:

[0102] The first module is used to obtain various influencing indicators that affect the air quality in the monitoring area through preset sensors, and obtain the target quality level of the monitoring area based on the various influencing indicators and through a preset algorithm model or network model.

[0103] The second module is used to control the fresh air fan to operate in the corresponding fresh air fan working state based on the preset working states of each different level and according to the target quality level.

[0104] Furthermore, in the first module, the target quality level of the monitoring area is obtained through the algorithm model, specifically including:

[0105] The first submodule is used to calculate the degree of membership of each influencing indicator to each level and the influence weight of each influencing indicator based on the preset number of different levels and each influencing indicator.

[0106] The second submodule is used to calculate the target quality level of the monitoring area based on a preset fuzzy operator according to the membership degree and influence weight of each influencing indicator to each level.

[0107] Furthermore, in the first module, the target quality level of the monitoring area is obtained through the network model, specifically including:

[0108] The third submodule is used to optimize and train the preset initial model using the historical indicator data of the monitoring area and the historical levels corresponding to the historical indicator data, so as to obtain a network model that has been optimized and trained and corresponds to the preset initial model.

[0109] The fourth submodule is used to input various influencing indicators into the network model for processing, and the network model outputs the target quality level corresponding to each influencing indicator.

[0110] Embodiment 3:

[0111] This embodiment provides an electronic device, such as Figure 5As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any method in Embodiment 1 is implemented.

[0112] Embodiment 4:

[0113] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any method in Embodiment 1.

[0114] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for controlling air quality in an underground space environment, characterized in that: The specific steps include: Acquire various influencing indicators that affect air quality in the monitoring area through preset sensors, and obtain the target quality level of the monitoring area based on the various influencing indicators and through a preset algorithm model or network model; Based on the preset working states of the fresh air fans at different levels, the fresh air fans are controlled to operate in the corresponding working states of the fresh air fans according to the target quality level; wherein the target quality level of the monitoring area is obtained through the algorithm model, specifically including: Based on the number of preset different levels and each of the influencing indicators, the degree of membership of each of the influencing indicators to each level and the influencing weight of each of the influencing indicators are calculated; According to the membership degree and influence weight of each influencing indicator to each level, the target quality level of the monitoring area is calculated based on a preset fuzzy operator; The target quality level of the monitoring area is obtained through the network model, specifically including: Optimizing and training the preset initial model using the historical indicator data of the monitoring area and the historical levels corresponding to the historical indicator data, to obtain the network model that has been optimized and trained and corresponds to the preset initial model; Each of the influencing indicators is input into the network model for processing, and the network model outputs a target quality level corresponding to each of the influencing indicators.

2. The air quality control method for underground space environment according to claim 1, characterized in that: The optimization training includes: introducing a momentum factor in the iterative training of the preset initial model to adjust the model weight of the preset initial model, and the model parameters include the model weight, which is specifically: in, Where Δw(k+1) represents the change in model weight at the k+1th iteration, η represents the momentum factor, 0<η<1, Δw(k) represents the change in model weight at the kth iteration, α represents the model learning rate of the current iteration, E(k) represents the value of the error function of the preset initial model at the kth iteration, E(k)=1 / 2Σ(A(k)-B(k)) 2 , A(k) represents the actual output value of the preset initial model at the kth iteration, and B(k) represents the expected output value of the preset initial model at the kth iteration. Represents the partial derivative of the error function with respect to the model weights.

3. The air quality control method for underground space environment according to claim 2, characterized in that: The model parameters include a model learning rate, which is specifically: Wherein, α(k+1) represents the model learning rate at the k+1th iteration, α(k) represents the model learning rate at the kth iteration, E(k) represents the value of the error function of the preset initial model at the kth iteration, and E(k+1) represents the value of the error function of the preset initial model at the k+1th iteration.

4. The air quality control method for underground space environment according to claim 2, characterized in that: The model weights are also modified by: In the formula, Δw(k+1) represents the change in the model weight at the k+1th iteration, Δw(k) represents the change in the model weight at the kth iteration, σ represents the increment factor, and σ>1, β represents the decrement factor, and β<1, Represents the partial derivative of the error function with respect to the model weights.

5. The air quality control method for underground space environment according to claim 2, characterized in that: The model weights are also modified by: Δw(k+1)=Δw(k)-(J T +μI) -1 J T e; Where Δw(k+1) represents the change in the model weight at the k+1th iteration, 0<η<1, Δw(k) represents the change in the model weight at the kth iteration, J T Represents the Jacobian matrix, where the elements are the partial derivatives of the error function with respect to the weights, μ represents the correlation coefficient, I represents the identity matrix, and e represents the error vector.

6. The air quality control method for underground space environment according to claim 1, characterized in that: The degree of membership of each of the influencing indicators to each level is specifically: When j = 1, When j = 2, 3, 4...n-1, When j = n, In the formula, j represents the serial number of each different level, j = 1, 2, 3...n, r ij represents the degree of membership of the i-th influencing indicator to the j-th level, x i represents the measured value of the i-th impact indicator, S ij It represents the standard value of the i-th impact indicator corresponding to level j.

7. The air quality control method for underground space environment according to claim 1, characterized in that: The impact weight of each of the impact indicators is specifically: In the formula, a i represents the influence weight of the i-th influencing indicator, x i represents the measured value of the i-th impact indicator, S i It represents the mean of the standard values ​​of the i-th influencing indicator in each level.

8. An air quality control system for an underground space environment, applied to an air quality control method for an underground space environment according to any one of claims 1 to 7, characterized in that: include: The first module is used to obtain various influencing indicators that affect the air quality in the monitoring area through preset sensors, and obtain the target quality level of the monitoring area based on each of the influencing indicators and through a preset algorithm model or network model; The second module is used to control the fresh air fan to operate in the corresponding fresh air fan working state based on the preset working states of each different level, and according to the target quality level; wherein the target quality level of the monitoring area is obtained by the algorithm model in the first module, specifically including: The first submodule is used to calculate the degree of membership of each influencing indicator to each level and the influence weight of each influencing indicator based on the number of preset different levels and each influencing indicator; The second submodule is used to calculate the target quality level of the monitoring area based on a preset fuzzy operator according to the membership degree and influence weight of each influencing indicator to each level; The first module obtains the target quality level of the monitoring area through the network model, specifically including: The third submodule is used to optimize and train the preset initial model using the historical indicator data of the monitoring area and the historical levels corresponding to the historical indicator data, so as to obtain the network model that has completed the optimization training and corresponds to the preset initial model; The fourth submodule is used to input each of the influencing indicators into the network model for processing, and the network model outputs a target quality level corresponding to each of the influencing indicators.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the method according to any one of claims 1 to 7 is implemented when the processor executes the computer program.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which enable a computer to execute the method of any one of claims 1-7.

Images