Intelligent parking lot management system and method based on multi-dimensional data

Abstract

The invention provides an intelligent parking lot management system and method based on multi-dimensional data, relates to the field of intelligent management, and aims to understand the air quality change trend of an underground parking lot by performing time sequence analysis on CO concentration values, PM2.5 concentration values and PM10 concentration values of the underground parking lot at multiple time points, and decide whether to start a ventilation system or not based on the air quality change trend. Therefore, the air quality change in the parking lot can be mastered more accurately, ventilation measures can be taken in time, it is ensured that the air quality is maintained at the safety level, unnecessary operation of a ventilation system can be reduced, and energy waste can be reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent management, and more specifically, to an intelligent parking lot management system and method based on multi-dimensional data. Background Art

[0002] An underground parking lot is usually a relatively closed space. Pollutants such as carbon monoxide (CO) and particulate matter (such as PM2.5 and PM10) emitted when vehicles enter and exit are not easy to diffuse and can easily accumulate to concentrations that are harmful to the human body in a short period of time. In order to effectively improve the air quality in underground parking lots, ventilation systems are generally installed in underground parking lots. Some existing underground parking lots may rely on fixed schedules or preset thresholds to turn on the ventilation system, which may lead to unnecessary energy waste or failure to improve air quality in a timely manner.

[0003] Therefore, an intelligent parking lot management solution based on multi-dimensional data is needed. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide an intelligent parking lot management system and method based on multi-dimensional data.

[0005] According to one aspect of the present application, there is provided an intelligent parking lot management system based on multidimensional data, which includes:

[0006] The underground parking lot monitoring data collection module is used to obtain the CO concentration value, PM2.5 concentration value and PM10 concentration value at multiple predetermined time points within a predetermined time period in the underground parking lot;

[0007] The underground parking lot monitoring data time series encoding module is used to perform a comprehensive time series analysis on the CO concentration values, PM2.5 concentration values ​​and PM10 concentration values ​​at the plurality of predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector;

[0008] The underground parking lot monitoring data time series correlation analysis module is used to perform time series correlation analysis on the parking lot environment parameter time series fusion feature vector to obtain the parking lot environment parameter time series correlation feature vector;

[0009] The ventilation system control result generation module is used to determine whether to turn on the ventilation system of the underground parking lot according to the information in the time series associated feature vector of the parking lot environmental parameters.

[0010] According to another aspect of the present application, a method for intelligent parking lot management based on multidimensional data is provided, which includes:

[0011] Obtain CO concentration values, PM2.5 concentration values, and PM10 concentration values ​​at multiple predetermined time points within a predetermined time period in the underground parking lot;

[0012] Performing a comprehensive time series analysis on the CO concentration values, PM2.5 concentration values, and PM10 concentration values ​​at the plurality of predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector;

[0013] Performing time series correlation analysis on the parking lot environment parameter time series fusion feature vector to obtain a parking lot environment parameter time series correlation feature vector;

[0014] Whether to turn on the ventilation system of the underground parking lot is determined according to the information in the time series associated feature vector of the parking lot environmental parameters.

[0015] Compared with the prior art, the intelligent parking lot management system and method based on multi-dimensional data provided by the present application can understand the trend of air quality changes in the underground parking lot by performing time series analysis on CO concentration values, PM2.5 concentration values, and PM10 concentration values ​​at multiple time points in the underground parking lot, and decide whether to turn on the ventilation system based on this. In this way, the changes in air quality in the parking lot can be more accurately grasped, ventilation measures can be taken in time to ensure that the air quality is maintained at a safe level, and unnecessary operation of the ventilation system can be reduced, which is conducive to reducing energy waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0017] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying 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 of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 4 is a system block diagram of an intelligent parking lot management system based on multi-dimensional data according to an embodiment of the present application.

[0019] Figure 2 The block diagram is a time series encoding module for underground parking lot monitoring data in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application.

[0020] Figure 3 4 is a block diagram of a CO concentration time series feature extraction unit in an intelligent parking lot management system based on multi-dimensional data according to an embodiment of the present application.

[0021] Figure 4 It is a block diagram of a solid particulate matter concentration time series feature extraction unit in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application.

[0022] Figure 5 It is a block diagram of a time series correlation analysis module for underground parking lot monitoring data in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application.

[0023] Figure 6 The block diagram is a module for generating ventilation system control results in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application.

[0024] Figure 7 It is a flowchart of an intelligent parking lot management method based on multi-dimensional data according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0026] As mentioned in the above background technology, in modern urban construction, underground parking lots, as an indispensable infrastructure, provide parking space for a large number of vehicles. However, underground parking lots are usually in a relatively closed environment, surrounded by buildings or soil layers, and only have limited entrances and exits connected to the outside world. When vehicles enter and exit, exhaust emissions become the main source of air pollution. Among them, carbon monoxide (CO), as a colorless and odorless but toxic gas, once inhaled by the human body, will bind to hemoglobin, reduce the oxygen carrying capacity of the blood, and pose a serious threat to human health. In addition, particulate matter emitted by vehicles, such as PM2.5 and PM10, should not be ignored. These tiny particles can penetrate into the human respiratory system and cause respiratory diseases, cardiovascular problems, etc.

[0027] Due to the closed nature of underground parking lots, these pollutants are difficult to diffuse naturally and tend to accumulate rapidly in a short period of time. If the concentration reaches a level that is harmful to the human body, it will not only affect the health of parking lot workers and car owners, but long-term exposure may also lead to more serious health risks.

[0028] In order to effectively improve the air quality of underground parking lots, ventilation systems have become indispensable facilities. Through mechanical ventilation equipment, fresh air is introduced into the parking lot and polluted air is discharged, thereby diluting and reducing the concentration of pollutants. However, some existing underground parking lots have certain limitations in the operation and management of ventilation systems. Some parking lots rely on fixed working hours to turn on the ventilation system, such as continuous operation during the day or a specific time period. This method does not take into account the actual vehicle flow and pollutant emissions. During periods with fewer vehicles, the continuous operation of the ventilation system will cause unnecessary energy waste. On the other hand, some parking lots use a preset threshold method to start the ventilation system when the pollutant concentration reaches a certain value. However, since the accumulation rate of pollutants may be relatively fast, this method may not be able to respond in time, resulting in the air quality in the parking lot having adverse effects on the human body before the concentration rises to the threshold. Therefore, an intelligent parking lot management solution based on multidimensional data is expected.

[0029] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also shown a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks provides new solutions and solutions for intelligent parking lot management based on multi-dimensional data.

[0030] Figure 1 : is a system block diagram of an intelligent parking lot management system based on multidimensional data according to an embodiment of the present application. Figure 1 As shown, in the intelligent parking lot management system 100 based on multi-dimensional data, it includes: an underground parking lot monitoring data acquisition module 110, which is used to obtain the CO concentration value, PM2.5 concentration value and PM10 concentration value at multiple predetermined time points within a predetermined time period in the underground parking lot; an underground parking lot monitoring data time series encoding module 120, which is used to perform a comprehensive time series analysis on the CO concentration value, PM2.5 concentration value and PM10 concentration value at the multiple predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector; an underground parking lot monitoring data time series correlation analysis module 130, which is used to perform a time series correlation analysis on the parking lot environmental parameter time series fusion feature vector to obtain a parking lot environmental parameter time series correlation feature vector; a ventilation system control result generation module 140, which is used to decide whether to turn on the ventilation system of the underground parking lot according to the information in the parking lot environmental parameter time series correlation feature vector.

[0031] In the embodiment of the present application, the underground parking lot monitoring data acquisition module 110 is used to obtain the CO concentration value, PM2.5 concentration value and PM10 concentration value at multiple predetermined time points in the underground parking lot within a predetermined time period. It should be understood that CO is one of the main components of automobile exhaust and is a toxic gas. In a relatively closed space such as an underground parking lot, vehicles frequently enter and exit, and a large amount of CO is emitted when the engine combustion is incomplete. When the human body inhales excessive CO, it will combine with hemoglobin, causing it to lose its oxygen-carrying capacity, leading to hypoxia of human tissues and organs, causing symptoms such as headaches, dizziness, nausea, vomiting, and even life-threatening in severe cases. Therefore, monitoring CO concentration is crucial to protecting the life, health and safety of people in underground parking lots. PM2.5 and PM10 are important components of atmospheric pollutants. PM2.5 refers to particles with an aerodynamic equivalent diameter of less than or equal to 2.5 microns in the ambient air, and PM10 refers to particles with an aerodynamic equivalent diameter of less than or equal to 10 microns in the ambient air. During vehicle driving, tire wear, exhaust emissions, etc. will produce particles. These particles can penetrate deep into the human respiratory system, and PM2.5 can even enter the alveoli and enter the blood circulation, causing respiratory diseases, cardiovascular diseases and other health problems. Moreover, they will reduce the air quality and visibility in the underground parking lot, affecting people's normal activities. Therefore, monitoring the concentration values ​​of PM2.5 and PM10 can comprehensively evaluate the air quality in the underground parking lot. Considering that the air quality in the underground parking lot is not static, it will change with factors such as the frequency and time of vehicle entry and exit. For example, during rush hour, vehicles frequently enter and exit, and the emission of pollutants will increase sharply in a short period of time, and the concentrations of CO, PM2.5 and PM10 will rise rapidly. During non-peak hours, the concentration may gradually decrease. By obtaining the index values ​​at multiple time points, this dynamic change process can be captured, and then the changing trend of the parking lot air quality can be understood, so as to make more reasonable ventilation system control decisions.

[0032] In the embodiment of the present application, the underground parking lot monitoring data time series encoding module 120 is used to perform a comprehensive time series analysis on the CO concentration values, PM2.5 concentration values ​​and PM10 concentration values ​​at the plurality of predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector. Specifically, Figure 2 FIG. 1 is a block diagram of a time series encoding module for underground parking lot monitoring data in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application. Figure 2As shown, the underground parking lot monitoring data time series encoding module 120 includes: a CO concentration time series feature extraction unit 121, which is used to perform time series feature extraction on the CO concentration values ​​at the multiple predetermined time points to obtain a CO concentration time series feature vector; a solid particle concentration time series feature extraction unit 122, which is used to perform time series feature extraction on the PM2.5 concentration values ​​and PM10 concentration values ​​at the multiple predetermined time points to obtain a solid particle concentration time series feature vector; a parking lot environment parameter time series feature fusion unit 123, which is used to fuse the CO concentration time series feature vector and the solid particle concentration time series feature vector to obtain the parking lot environment parameter time series fusion feature vector.

[0033] In the embodiment of the present application, the CO concentration time series feature extraction unit 121 is used to extract the time series features of the CO concentration values ​​at the plurality of predetermined time points to obtain a CO concentration time series feature vector. Specifically, Figure 3 FIG. 1 is a block diagram of a CO concentration time series feature extraction unit in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application. Figure 3 As shown, the CO concentration time series feature extraction unit 121 includes: a CO concentration value data structured arrangement subunit 1211, used to arrange the CO concentration values ​​of the multiple predetermined time points according to the time dimension to obtain a CO concentration time series input vector; a CO concentration time series feature generation subunit 1212, used to pass the CO concentration time series input vector through a CO concentration time series change capturer to obtain the CO concentration time series feature vector.

[0034] In an embodiment of the present application, the CO concentration value data structured arrangement subunit 1211 is used to arrange the CO concentration values ​​of the multiple predetermined time points according to the time dimension to obtain a CO concentration time series input vector. It should be understood that the carbon monoxide (CO) concentration in the underground parking lot will change dynamically over time. For example, the concentration will increase during peak hours when vehicles enter and exit, and decrease during non-peak hours. In order to clearly understand the changes in CO concentration over time, such as whether the CO concentration is rising, falling, or remaining stable, it is necessary to arrange the CO concentration values ​​of the multiple predetermined time points according to the time dimension. The arranged CO concentration time series input vector can facilitate the model to capture the concentration change trend of CO, and thus better understand the air quality in the parking lot. In addition, in computer processing, data in vector form is easier to perform batch operations and management, and can also efficiently use storage space.

[0035] In an embodiment of the present application, the CO concentration time series feature generation subunit 1212 is used to pass the CO concentration time series input vector through the CO concentration time series change capturer to obtain the CO concentration time series feature vector. It should be understood that the original CO concentration time series input vector contains a large number of time series data points. Although these data can reflect the change of concentration over time, they contain a lot of redundant information. In order to focus on the truly valuable parts of the data, such as the trend of concentration change, periodic characteristics, mutation points, etc., it is necessary to extract the time series features of the CO concentration time series input vector. Through the time series feature extraction, the key information in the CO concentration time series input vector can be extracted, which facilitates more accurate judgment of the change pattern of the air quality of the underground parking lot. Specifically, the CO concentration time series input vector is input into the CO concentration time series change capturer for feature extraction processing. In particular, the CO concentration time series change capturer is a convolutional neural network model using a one-dimensional convolution kernel. Those of ordinary skill in the art should know that convolutional neural networks have the ability to automatically extract features. For a one-dimensional CO concentration time series input vector, the one-dimensional convolution kernel can slide along the time dimension and automatically learn local features in the data. For example, it can detect features such as short-term fluctuations, long-term trends, and periodic changes in concentration without manually designing feature extraction rules, which greatly reduces the impact of human factors on feature extraction. And when processing time series data, the one-dimensional convolution kernel can well preserve the time order of the data, which helps to better understand the dynamic changes in CO concentration.

[0036] In the embodiment of the present application, the solid particle concentration time series feature extraction unit 122 is used to extract the time series features of the PM2.5 concentration values ​​and PM10 concentration values ​​at the plurality of predetermined time points to obtain a solid particle concentration time series feature vector. Specifically, Figure 4 FIG. 1 is a block diagram of a solid particulate matter concentration time series feature extraction unit in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application. Figure 4 As shown, the solid particle concentration time series feature extraction unit 122 includes: a solid particle concentration data structuring subunit 1221, which is used to arrange the PM2.5 concentration values ​​and PM10 concentration values ​​of the multiple predetermined time points according to the time dimension to obtain a solid particle concentration time series input matrix; a solid particle concentration time series feature generation subunit 1222, which is used to pass the solid particle concentration time series input matrix through a solid particle concentration time series change capturer to obtain the solid particle concentration time series feature vector.

[0037] In an embodiment of the present application, the solid particle concentration data structured arrangement subunit 1221 is used to arrange the PM2.5 concentration values ​​and PM10 concentration values ​​at the multiple predetermined time points according to the time dimension to obtain a solid particle concentration time series input matrix. It should be understood that PM2.5 and PM10 are two important solid particle indicators in underground parking lot air quality monitoring. Their concentration changes are interrelated in time and affect the air quality together. In order to simultaneously consider the changes of these two indicators at different time points and conduct a comprehensive analysis, it is necessary to arrange the PM2.5 concentration values ​​and PM10 concentration values ​​at the multiple predetermined time points according to the time dimension in the present application. In the solid particle concentration time series input matrix obtained by arrangement, each row of data contains the concentration values ​​of PM2.5 and PM10 at the same time point, and each column of data represents the concentration of particles of different solid sizes.

[0038] In an embodiment of the present application, the solid particle concentration time series feature generation subunit 1222 is used to pass the solid particle concentration time series input matrix through the solid particle concentration time series change capturer to obtain the solid particle concentration time series feature vector. It should be understood that the solid particle concentration time series input matrix contains the concentration information of PM2.5 and PM10 at multiple time points, the data volume is large and may contain a lot of redundant information. In order to mine the most critical information in the solid particle concentration time series data, such as the trend of the change of the concentration of the two particles (whether it is rising, falling or fluctuating), the periodic characteristics (whether the concentration peak appears in a specific time period), the correlation between the concentration changes of the two (whether it is synchronous change or there is a lag relationship), etc., it is necessary to perform time series feature extraction processing on the solid particle concentration time series input matrix. Through the time series feature extraction processing, the key features in the original time series data can be extracted, and these features can more accurately reflect the essential situation of solid particle pollution in the underground parking lot. Specifically, the solid particle concentration time series input matrix is ​​input into the solid particle concentration time series change capturer for feature extraction processing. In particular, the solid particle concentration time series change capturer is a convolutional neural network model in which adjacent layers use mutually transposed convolution kernels. It should be noted that ordinary convolutional neural networks mainly perform feature extraction through unidirectional convolution kernels, focusing on capturing feature changes in data in one direction. Here, a convolutional neural network model in which adjacent layers use mutually transposed convolution kernels can achieve bidirectional feature extraction. In detail, when processing the solid particle concentration time series input matrix, the convolution kernel in one direction can capture the concentration change characteristics along the time series, such as the rising or falling trend of the concentration; while the transposed convolution kernel can extract features from another perspective, which may focus on the correlation between concentration changes at different time points or the mutual influence between PM2.5 and PM10 concentration changes. This bidirectional feature extraction can more comprehensively mine the information in the matrix, enhance the feature expression capability, and enable the extracted time series feature vector to more accurately reflect the complex changes in the solid particle concentration.

[0039] In an embodiment of the present application, the parking lot environmental parameter time series feature fusion unit 123 is used to fuse the CO concentration time series feature vector and the solid particle concentration time series feature vector to obtain the parking lot environmental parameter time series fusion feature vector. It should be understood that in the underground parking environment, carbon monoxide (CO) and solid particulate matter (PM2.5 and PM10) are important air pollutants. They come from a variety of vehicle-related activities such as vehicle exhaust emissions and tire wear. Considering the CO concentration or solid particle concentration alone can only provide a one-sided understanding of the air quality. For example, there may be a situation where the CO concentration is low but the solid particle concentration is very high, or vice versa. In order to comprehensively evaluate the air quality of the underground parking lot from multiple dimensions, it is necessary to fuse the CO concentration time series feature vector and the solid particle concentration time series feature vector in the present application. The parking lot environmental parameter time series fusion feature vector obtained after fusion can provide a comprehensive air quality index feature, which can more accurately reflect the changes in air quality in the underground parking lot.

[0040] In the embodiment of the present application, the underground parking lot monitoring data time series correlation analysis module 130 is used to perform time series correlation analysis on the parking lot environment parameter time series fusion feature vector to obtain the parking lot environment parameter time series correlation feature vector. Specifically, Figure 5 FIG. 1 is a block diagram of a time series correlation analysis module for underground parking lot monitoring data in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application. Figure 5 As shown, the underground parking lot monitoring data time series correlation analysis module 130 includes: a feature vector segmentation unit 131, used to segment the parking lot environment parameter time series fusion feature vector to obtain multiple parking lot environment parameter time series fusion feature sub-vectors; a time series correlation feature generation unit 132, used to pass the multiple parking lot environment parameter time series fusion feature sub-vectors through a time series change correlation capturer to obtain the parking lot environment parameter time series correlation feature vector.

[0041] In an embodiment of the present application, the feature vector segmentation unit 131 is used to segment the parking lot environmental parameter time series fusion feature vector to obtain multiple parking lot environmental parameter time series fusion feature sub-vectors. It should be understood that the parking lot environmental parameter time series fusion feature vector contains comprehensive time series information of carbon monoxide (CO) concentration and solid particulate matter concentration. However, the time series represented by the entire vector may be long, and during this long period of time, the change in air quality may have different stages and local characteristics. By segmenting the parking lot environmental parameter time series fusion feature vector, it can be divided into multiple sub-vectors, each of which can focus on a shorter time interval, so that the local change trend of air quality in these shorter time intervals can be captured more carefully.

[0042] In an embodiment of the present application, the time series correlation feature generating unit 132 is used to pass the time series fusion feature sub-vectors of the multiple parking lot environmental parameters through a time series change correlation capturer to obtain the parking lot environmental parameter time series correlation feature vector. In particular, the time series change correlation capturer is a bidirectional long short-term memory neural network model. It should be understood that the bidirectional long short-term memory neural network model (Bi-LSTM) is composed of a forward LSTM and a backward LSTM. The forward LSTM can process the sequence data from front to back in chronological order to capture the positive time dependency of the data; while the backward LSTM processes the sequence data from back to front, and it can capture the reverse time dependency. Considering that the time series fusion feature sub-vectors of multiple parking lot environmental parameters represent the comprehensive characteristics of air quality in different time intervals, there is a time series correlation between these sub-vectors, which is crucial for a comprehensive understanding of the dynamic changes in the air quality of underground parking lots. By inputting the time series fusion feature sub-vectors of the multiple parking lot environmental parameters into the time series change correlation capturer (Bi-LSTM), the order relationship and mutual influence between these sub-vectors can be mined. For example, the changes in air quality during peak hours recorded in a sub-vector may have a lasting impact on the air quality during non-peak hours represented in subsequent sub-vectors. For example, high concentrations of pollutants generated during peak hours may gradually diffuse and dilute over a period of time. Bi-LSTM can capture this cross-sub-vector temporal correlation very well. The generated parking lot environmental parameter temporal correlation feature vector not only contains the overall air quality change trend information of the parking lot, but also reflects the mutual influence of air quality between different time periods. Making ventilation control decisions based on this information can make the decision more reasonable and scientific.

[0043] In the embodiment of the present application, the ventilation system control result generation module 140 is used to determine whether to turn on the ventilation system of the underground parking lot according to the information in the parking lot environmental parameter time series associated feature vector. Specifically, Figure 6 FIG. 1 is a block diagram of a ventilation system control result generation module in a multi-dimensional data-based intelligent parking lot management system according to an embodiment of the present application. Figure 6 As shown, the ventilation system control result generation module 140 includes: a parking lot environment parameter time series association feature optimization unit 141, which is used to perform feature matching optimization on the parking lot environment parameter time series association feature vector based on target domain distribution characteristic adaptation to obtain an optimized parking lot environment parameter time series association feature vector; a parking lot environment parameter time series association feature parsing unit 142, which is used to pass the optimized parking lot environment parameter time series association feature vector through a ventilation control classifier to obtain a ventilation control classification result, and the ventilation control classification result is used to indicate whether to turn on the ventilation system of the underground parking lot.

[0044] In an embodiment of the present application, the parking lot environmental parameter time series correlation feature optimization unit 141 is used to perform feature matching optimization based on the adaptation of the target domain distribution characteristics on the parking lot environmental parameter time series correlation feature vector to obtain an optimized parking lot environmental parameter time series correlation feature vector. In particular, it is considered that different underground parking lots may be in different geographical locations and surrounding environments. For example, a parking lot located near a busy main road may have a higher concentration of CO, PM2.5 and PM10 as a whole, and the concentration changes more frequently and violently due to the exhaust emissions of passing vehicles; while a parking lot located in a relatively remote area with low traffic volume has a relatively low pollutant concentration and changes more slowly. This results in different distributions of the concentration values ​​of each pollutant at the same predetermined time point in different parking lots, which in turn affects the distribution of the time series vector formed by the arrangement according to the time dimension. In traditional methods, when training a model, it is usually assumed that the training data and the target domain data have the same or similar distribution. However, as mentioned above, there are significant differences in the distribution of environmental parameter characteristics of different parking lots, which can lead to misjudgment of the ventilation control results based on the parking lot environmental parameter time series correlation feature vector. Therefore, in the technical solution of the present application, it is necessary to perform feature matching optimization on the parking lot environment parameter time series association feature vector based on the adaptation of the target domain distribution characteristics to obtain an optimized parking lot environment parameter time series association feature vector.

[0045] Specifically, in an embodiment of the present application, the parking lot environment parameter time series association feature optimization unit is used to: map the parking lot environment parameter time series association feature vector to the intrinsic decomposition space to obtain the parking lot environment parameter time series association intrinsic decomposition feature vector; extract the maximum eigenvalue and the minimum eigenvalue of the parking lot environment parameter time series association intrinsic decomposition feature vector; calculate the difference between the maximum eigenvalue and the minimum eigenvalue as the parking lot environment monitoring target domain edge anchor description operator; extract the mean and standard deviation of the parking lot environment parameter time series association intrinsic decomposition feature vector, and divide the mean by the standard deviation to obtain the parking lot environment monitoring optimization direction description operator; based on the parking lot environment monitoring target domain edge anchor description operator and the parking lot environment monitoring optimization direction description operator, match and optimize the parking lot environment parameter time series association feature vector to obtain the optimized parking lot environment parameter time series association feature vector.

[0046] In the embodiment of the present application, specifically, the parking lot environment parameter time series association feature vector is subjected to feature matching optimization based on target domain distribution characteristic adaptation to obtain an optimized parking lot environment parameter time series association feature vector, including: processing the parking lot environment parameter time series association feature vector according to the following formula to obtain the optimized parking lot environment parameter time series association feature vector; wherein the formula is:

[0047]

[0048] α=v max -v min

[0049]

[0050] Wherein, V1 represents the parking lot environment parameter time series correlation feature vector, PCA(V1) represents mapping V1 to the eigendecomposition space, U1 is the sequence of parking lot environment parameter time series correlation eigendecomposition vectors, Λ1 is the parking lot environment parameter time series correlation diagonal matrix, U1 T is the transpose of U1, v 11 、v 12 、v 1m are the first, second and mth parking lot environment parameter time series associated intrinsic decomposition vectors of the sequence of parking lot environment parameter time series associated intrinsic decomposition vectors, λ 11 , 1m are the eigenvalues ​​of the first and mth positions of the parking environment parameter time series correlation diagonal matrix, V represents the eigenvalue vector of the parking environment parameter time series correlation intrinsic decomposition, v max and v min They represent the maximum eigenvalue and minimum eigenvalue of the eigenvalue decomposition eigenvector of the parking lot environment parameter time series association, α represents the edge anchor description operator of the parking lot environment monitoring target domain, μ and σ represent the mean and standard deviation of the eigenvalue decomposition eigenvector of the parking lot environment parameter time series association, τ represents the parking lot environment monitoring optimization direction description operator, ⊙ and They represent point-by-point addition, point-by-point multiplication, and point-by-point subtraction, respectively. exp represents the natural exponential function with the natural constant e as the base. V1 ⊙-1 represents the calculation of the inverse of each eigenvalue of the parking lot environment parameter time series associated feature vector, and V' represents the optimized parking lot environment parameter time series associated feature vector.

[0051] In view of the above technical problems, in the technical solution of the present application, the parking lot environmental parameter time series correlation feature vector is optimized based on the feature matching adapted to the target domain distribution characteristics. The process first maps the parking lot environmental parameter time series correlation feature vector to the eigendecomposition space. With the help of eigendecomposition, the original high-dimensional parking lot environmental parameter time series correlation feature vector is projected into a low-dimensional space composed of an orthogonal basis. The direction in this low-dimensional space is defined by the eigenvector, and its corresponding eigenvalue describes the variance or intensity distribution of the data in different directions. The process of projecting to the eigendecomposition space not only removes the redundant noise in the original parking lot environmental parameter time series correlation characteristics by dimensionality reduction, but also maps the original vector of complex distribution to a more physically intuitive decomposition dimension, making the structured information of the feature clearer, which can make it easier for the model to understand the parking lot environment change mechanism behind the feature data, so as to more accurately grasp the changing law of pollutant concentration.

[0052] After the mapping is completed, it is necessary to further extract the maximum eigenvalue and minimum eigenvalue of the eigenvalue decomposition eigenvector of the time series correlation of the parking environment parameters. The direction corresponding to the maximum eigenvalue is the direction in which the data changes most significantly, that is, the main direction. In the context of the eigenvalue decomposition eigenvector of the time series correlation of the parking environment parameters, this main direction represents the combination of factors that have the most critical impact on the parking environment. The minimum eigenvalue is usually associated with noise or data error, and the direction corresponding to it is the direction in which the data changes the weakest. In parking environment monitoring, factors such as the accuracy limitations of the measuring equipment and random interference in the environment will introduce noise. These noises appear in the data as some small and irregular changes, and the direction corresponding to the minimum eigenvalue is the embodiment of these weak changes. The maximum and minimum eigenvalues ​​are important representations of the spatial morphology of the data. In the analysis and optimization process, they provide a mathematical description of the global distribution characteristics of the characteristics of the parking environment parameters. At the same time, the distribution pattern of the eigenvalues ​​also implies the complexity of the parking environment data in the target domain. When the maximum eigenvalue is much larger than other eigenvalues, it indicates that the feature has an obvious main axis direction in the space, which means that the changes in the parking environment are mainly dominated by several key factors. The internal structure of the data is relatively simple, and analysis and prediction are relatively easy. If the eigenvalue distribution is relatively uniform and there is no obvious dominant maximum eigenvalue, this may imply that there are multiple factors interacting with each other in the parking environment and the influence is relatively balanced, and the data has higher complexity or diversity.

[0053] In order to construct further feature optimization indicators, it is necessary to calculate the difference between the maximum eigenvalue and the minimum eigenvalue and define it as the edge anchoring descriptor of the parking lot environmental monitoring target domain. The eigenvalue difference quantifies the distribution range and span of the parking lot environmental parameter characteristics in the target domain from a geometric perspective. A larger difference means that the changes in the parking lot environmental parameters in different directions (defined by the eigenvectors) are significantly different, reflecting that the environment may be in an unstable state. For example, in a parking lot with large traffic volume changes and poor ventilation, the pollutant concentration may rise sharply in some periods and be relatively stable in other periods, which will lead to a large difference between the maximum and minimum eigenvalues. At this time, the larger difference indicates that the environmental parameters fluctuate greatly, and the ventilation system may need to be turned on more frequently to maintain stable air quality. On the contrary, the smaller difference indicates that the environmental parameters change more steadily and the frequency of ventilation system opening may be relatively low. In other words, the difference can be used as a boundary indicator for the regular fluctuations of parking lot environmental parameters. In daily operations, each parking lot has its relatively stable range of environmental parameter fluctuations. The eigenvalue difference calculated through long-term monitoring can help determine this regular range. When the characteristic value difference corresponding to the environmental parameters monitored in real time is close to or exceeds this normal range, it indicates that the environment may be abnormal and the ventilation system needs to be turned on for intervention.

[0054] At the same time, it is also necessary to extract the mean and standard deviation from the eigenvalue decomposition eigenvector of the parking lot environmental parameter time series correlation, and calculate their ratio to obtain the parking lot environmental monitoring optimization direction description operator. It should be understood that in parking lot environmental monitoring, different environmental parameters (such as CO concentration, PM2.5 concentration, and PM10 concentration) may have different data scales. For example, the CO concentration may be at the ppm level, while the PM2.5 concentration may be at μg / m 3level. When analyzing these data of different scales, some features may dominate the analysis results, while the role of other features is ignored. By calculating the parking lot environment monitoring optimization direction description operator (the ratio of mean to standard deviation), data of different scales can be normalized. This ratio eliminates the influence of data scale, so that all features are considered under a unified standard. For example, whether it is CO concentration or PM2.5 concentration, the ratio of their mean to standard deviation is at the same order of magnitude, so that in the subsequent feature optimization process, the features of different parameters can be reasonably considered and will not be biased due to differences in data scale. The operator can also be understood as an adaptive adjustment rule for direction selection. Specifically, the environment of the parking lot is complex and changeable. Different parking lots have different environmental characteristics. Even the same parking lot may have different environmental conditions in different time periods. The ratio of mean to standard deviation can adaptively adjust the direction of feature optimization according to the actual distribution of the data. For example, during periods of heavy traffic, the distribution of pollutant concentrations may change, and the mean and standard deviation will change accordingly, resulting in changes in the ratio. The model can automatically adjust the focus direction of the features according to this change, so that the optimization results are more in line with the actual environmental characteristics of the current parking lot.

[0055] After the construction of the above key operators is completed, the final matching optimization operation of the parking lot environmental parameter time series correlation feature vector uses the parking lot environmental monitoring target domain edge anchoring description operator and the parking lot environmental monitoring optimization direction description operator to improve the resolution of the feature vector and obtain the optimized representation. It should be understood that in parking lot environmental monitoring, focusing on the edge attributes or directional attributes of feature distribution alone may cause problems. If only focusing on edge attributes, it may overfit specific boundary conditions in the target domain and ignore the generality of overall features, causing the model to lose accuracy when facing slightly changing data; if only focusing on directional attributes, it may be biased towards certain dominant directions, ignoring the subtle but important changes in data in other directions, resulting in deviations. Matching optimization based on the parking lot environmental monitoring target domain edge anchoring description operator and the parking lot environmental monitoring optimization direction description operator, as a multi-objective optimization strategy, can balance these two attributes at the same time. For example, when determining the ventilation system control strategy, the threshold is not set based solely on the extreme conditions of environmental parameter changes (marginal attributes), nor is it set based solely on common change trends (directional attributes). Instead, a comprehensive consideration of both is used to make the model more comprehensive and robust, avoiding overfitting or deviation caused by preference for a certain goal. This can effectively improve the accuracy and efficiency of parking lot ventilation system control.

[0056] In the embodiment of the present application, the parking lot environmental parameter time series correlation feature parsing unit 142 is used to pass the optimized parking lot environmental parameter time series correlation feature vector through the ventilation control classifier to obtain a ventilation control classification result, and the ventilation control classification result is used to indicate whether to turn on the ventilation system of the underground parking lot. It should be understood that the optimized parking lot environmental parameter time series correlation feature vector contains rich air quality change information, but it is still a representation at the data level. In order to actually control the ventilation system, it is necessary to convert these data into clear decisions, that is, whether to turn on the ventilation system, and it is necessary to input the optimized parking lot environmental parameter time series correlation feature vector into the ventilation control classifier for parsing. The ventilation control classifier is a machine learning model, and its main function is to classify according to the input optimized parking lot environmental parameter time series correlation feature vector. This classification process is actually to divide different air quality conditions and judge whether the current air quality reaches the level where the ventilation system needs to be turned on. The ventilation control classification result obtained by classification directly determines whether the ventilation system of the underground parking lot is turned on or off. If the classification result shows that the air quality has dropped to a harmful level or has a tendency to deteriorate, then the ventilation system will be turned on to improve the air quality and protect the health of people in the parking lot. Conversely, if the air quality is good, the classification result will indicate that the ventilation system should not be turned on to avoid unnecessary energy consumption.

[0057] In summary, the multi-dimensional data-based intelligent parking lot management system 100 according to the embodiment of the present application is explained, which performs time series analysis on the CO concentration values, PM2.5 concentration values, and PM10 concentration values ​​at multiple time points in the underground parking lot to understand the trend of air quality changes in the underground parking lot, and decides whether to turn on the ventilation system based on this. In this way, the changes in air quality in the parking lot can be more accurately grasped, ventilation measures can be taken in time, and the air quality can be maintained at a safe level, and unnecessary operation of the ventilation system can be reduced, which is conducive to reducing energy waste.

[0058] Figure 7 FIG. 1 is a flow chart of a smart parking lot management method based on multidimensional data according to an embodiment of the present application. Figure 7As shown, in the intelligent parking lot management method based on multi-dimensional data, it includes: S110, obtaining CO concentration values, PM2.5 concentration values ​​and PM10 concentration values ​​at multiple predetermined time points within a predetermined time period in the underground parking lot; S120, performing a comprehensive time series analysis on the CO concentration values, PM2.5 concentration values ​​and PM10 concentration values ​​at the multiple predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector; S130, performing a time series correlation analysis on the parking lot environmental parameter time series fusion feature vector to obtain a parking lot environmental parameter time series correlation feature vector; S140, deciding whether to turn on the ventilation system of the underground parking lot according to the information in the parking lot environmental parameter time series correlation feature vector.

[0059] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned intelligent parking lot management method based on multi-dimensional data have been referred to above. Figures 1 to 6 The invention has been introduced in detail in the description of the intelligent parking lot management system based on multi-dimensional data, and therefore, its repeated description will be omitted.

[0060] In summary, the intelligent parking lot management method based on multi-dimensional data according to the embodiment of the present application is explained, which performs time series analysis on the CO concentration value, PM2.5 concentration value and PM10 concentration value at multiple time points in the underground parking lot to understand the trend of air quality changes in the underground parking lot, and decides whether to turn on the ventilation system based on this. In this way, the changes in air quality in the parking lot can be more accurately grasped, ventilation measures can be taken in time to ensure that the air quality is maintained at a safe level, and unnecessary operation of the ventilation system can be reduced, which is conducive to reducing energy waste.

Claims

1. An intelligent parking lot management system based on multidimensional data, characterized in that: include: The underground parking lot monitoring data collection module is used to obtain the CO concentration value, PM2.5 concentration value and PM10 concentration value at multiple predetermined time points within a predetermined time period in the underground parking lot; The underground parking lot monitoring data time series encoding module is used to perform a comprehensive time series analysis on the CO concentration values, PM2.5 concentration values ​​and PM10 concentration values ​​at the plurality of predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector; The underground parking lot monitoring data time series correlation analysis module is used to perform time series correlation analysis on the parking lot environment parameter time series fusion feature vector to obtain the parking lot environment parameter time series correlation feature vector; The ventilation system control result generation module is used to determine whether to turn on the ventilation system of the underground parking lot according to the information in the time series associated feature vector of the parking lot environmental parameters.

2. The intelligent parking lot management system based on multidimensional data according to claim 1 is characterized in that: The underground parking lot monitoring data time series encoding module includes: A CO concentration time series feature extraction unit, configured to extract time series features of the CO concentration values ​​at the plurality of predetermined time points to obtain a CO concentration time series feature vector; A solid particle concentration time series feature extraction unit, used for performing time series feature extraction on the PM2.5 concentration values ​​and PM10 concentration values ​​at the plurality of predetermined time points to obtain a solid particle concentration time series feature vector; The parking lot environmental parameter time series feature fusion unit is used to fuse the CO concentration time series feature vector and the solid particulate matter concentration time series feature vector to obtain the parking lot environmental parameter time series fusion feature vector.

3. The intelligent parking lot management system based on multidimensional data according to claim 2 is characterized in that: The CO concentration time series feature extraction unit comprises: A CO concentration value data structured arrangement subunit, configured to arrange the CO concentration values ​​at the plurality of predetermined time points according to a time dimension to obtain a CO concentration time series input vector; The CO concentration time series feature generating subunit is used to pass the CO concentration time series input vector through the CO concentration time series change capturer to obtain the CO concentration time series feature vector.

4. The intelligent parking lot management system based on multidimensional data according to claim 3 is characterized in that: The solid particle concentration time series feature extraction unit comprises: A solid particulate matter concentration data structured arrangement subunit, used for arranging the PM2.5 concentration values ​​and PM10 concentration values ​​at the plurality of predetermined time points according to the time dimension to obtain a solid particulate matter concentration time series input matrix; The solid particle concentration time series characteristic generating subunit is used to pass the solid particle concentration time series input matrix through the solid particle concentration time series change capture device to obtain the solid particle concentration time series characteristic vector.

5. The intelligent parking lot management system based on multi-dimensional data according to claim 4 is characterized in that: The underground parking lot monitoring data time series correlation analysis module includes: A feature vector segmentation unit, used for segmenting the parking lot environment parameter time series fusion feature vector to obtain a plurality of parking lot environment parameter time series fusion feature sub-vectors; The time series correlation feature generating unit is used to obtain the parking lot environment parameter time series correlation feature vector by passing the multiple parking lot environment parameter time series fusion feature sub-vectors through a time series change correlation capturer.

6. The intelligent parking lot management system based on multi-dimensional data according to claim 5 is characterized in that: The CO concentration time series change capturer is a convolutional neural network model using a one-dimensional convolution kernel, the solid particulate matter concentration time series change capturer is a convolutional neural network model in which adjacent layers use mutually transposed convolution kernels, and the time series change association capturer is a bidirectional long short-term memory neural network model.

7. The intelligent parking lot management system based on multi-dimensional data according to claim 6 is characterized in that: The ventilation system control result generating module comprises: A parking lot environment parameter time series correlation feature optimization unit, used for performing feature matching optimization based on target domain distribution characteristic adaptation on the parking lot environment parameter time series correlation feature vector to obtain an optimized parking lot environment parameter time series correlation feature vector; The parking lot environmental parameter time series correlation feature analysis unit is used to pass the optimized parking lot environmental parameter time series correlation feature vector through a ventilation control classifier to obtain a ventilation control classification result, and the ventilation control classification result is used to indicate whether to turn on the ventilation system of the underground parking lot.

8. The intelligent parking lot management system based on multi-dimensional data according to claim 7 is characterized in that: The parking lot environment parameter time series correlation feature optimization unit is used to: Mapping the parking lot environment parameter time series correlation feature vector to an eigendecomposition space to obtain a parking lot environment parameter time series correlation eigendecomposition feature vector; Extracting the maximum eigenvalue and the minimum eigenvalue of the parking lot environment parameter time series associated eigendecomposition eigenvector; Calculate the difference between the maximum eigenvalue and the minimum eigenvalue as an edge anchoring description operator for parking lot environment monitoring target domain; Extracting the mean and standard deviation of the parking lot environment parameter time series associated intrinsic decomposition eigenvector, and dividing the mean by the standard deviation to obtain a parking lot environment monitoring optimization direction description operator; Based on the parking lot environment monitoring target domain edge anchoring description operator and the parking lot environment monitoring optimization direction description operator, the parking lot environment parameter time series association feature vector is matched and optimized to obtain the optimized parking lot environment parameter time series association feature vector.

9. An intelligent parking lot management method based on multidimensional data, characterized in that: include: Obtain CO concentration values, PM2.5 concentration values, and PM10 concentration values ​​at multiple predetermined time points within a predetermined time period in the underground parking lot; Performing a comprehensive time series analysis on the CO concentration values, PM2.5 concentration values, and PM10 concentration values ​​at the plurality of predetermined time points to obtain a parking lot environmental parameter time series fusion feature vector; Performing time series correlation analysis on the parking lot environment parameter time series fusion feature vector to obtain a parking lot environment parameter time series correlation feature vector; Whether to turn on the ventilation system of the underground parking lot is determined according to the information in the time series associated feature vector of the parking lot environmental parameters.

10. The intelligent parking lot management method based on multi-dimensional data according to claim 9, characterized in that: Determining whether to turn on the ventilation system of the underground parking lot according to information in the time series correlation feature vector of the parking lot environmental parameter includes: Performing feature matching optimization based on target domain distribution characteristic adaptation on the parking lot environment parameter time series association feature vector to obtain an optimized parking lot environment parameter time series association feature vector; The optimized parking lot environmental parameter time series associated feature vector is passed through a ventilation control classifier to obtain a ventilation control classification result, and the ventilation control classification result is used to indicate whether to turn on the ventilation system of the underground parking lot.

Images