Parking equipment intelligent supervision method and system based on internet of things

Abstract

The application provides a kind of parking equipment wisdom supervision method and system based on Internet of Things, first acquire the multi-source monitoring data set of mechanical parking equipment that Internet of Things sensor network real-time collection, then carry out feature extraction to multi-source monitoring data set, generate structure stability features reflecting key component geometric shape time consistency and action fluency features reflecting execution action timing convergence continuity, then structure stability features and action fluency features are input into pre-training anomaly recognition model for anomaly evaluation, generate evaluation results containing abnormal existence identification and associated component information, then extract historical feature sequence comparison according to abnormal associated component information, determine abnormal influence range, finally generate supervision instruction containing maintenance priority and component identification based on abnormal influence range and comparison result and send to equipment management system to trigger maintenance response, realize the accurate supervision of mechanical parking equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Things, in particular to a parking equipment intelligent supervision method and system based on Internet of Things. BACKGROUND

[0002] Mechanical parking equipment, as an efficient parking solution, is widely used in various parking lots. However, mechanical parking equipment has a complex structure, including lifting mechanism, horizontal moving mechanism and locking mechanism and other key components. The above components are easily affected by factors such as wear and fatigue during long-term operation, resulting in equipment failure.

[0003] At present, the supervision of mechanical parking equipment mainly relies on manual regular inspection and simple sensor monitoring. Manual inspection is not only inefficient, but also difficult to grasp the running state of the equipment in real time, and cannot timely find potential fault hidden dangers. Although simple sensor monitoring can obtain some equipment operation data, these data are often isolated, lack effective integration and analysis, and it is difficult to comprehensively and accurately evaluate the mechanical properties and state of the equipment. In addition, the existing supervision method also has deficiencies in determining the fault influence range and formulating maintenance strategy, which cannot accurately maintain the equipment according to the actual running state of the equipment, resulting in high equipment maintenance cost and low efficiency, and even may cause safety accidents due to equipment failure. SUMMARY

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present application, the present application embodiment provides a parking equipment intelligent supervision method based on Internet of Things, which comprises:

[0005] obtaining a multi-source monitoring data set of mechanical parking equipment collected by an Internet of Things sensor network in real time, the multi-source monitoring data set comprising lifting mechanism displacement data, horizontal moving mechanism power data and locking mechanism state data in a continuous time period;

[0006] performing feature extraction on the multi-source monitoring data set to generate structure stability features and action fluency features of equipment mechanical properties, the structure stability features reflecting the time consistency of the geometric form of key components, and the action fluency features reflecting the continuity of action timing connection;

[0007] inputting the structure stability features and the action fluency features into a pre-trained anomaly recognition model for anomaly evaluation to generate an equipment state evaluation result containing anomaly existence identification and anomaly associated component information;

[0008] According to the anomaly associated component information in the equipment state evaluation result, the feature sequence of the corresponding anomaly associated component in the historical period is extracted and time sequence comparison analysis is performed to determine the abnormal influence range in the mechanical parking equipment.

[0009] Based on the comparison result of the abnormal influence range and the historical feature sequence, a supervision instruction containing a maintenance priority and a component identifier is generated, and the supervision instruction is sent to a device management system to trigger a maintenance response.

[0010] In still another aspect, the embodiment of the present application also provides a parking equipment intelligent supervision system based on Internet of Things, which comprises a processor and a machine readable storage medium, the machine readable storage medium is connected with the processor, the machine readable storage medium is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the machine readable storage medium to realize the above-mentioned method.

[0011] Based on the above aspects, the embodiment of the present application can obtain a mechanical parking equipment multi-source monitoring data set collected by an Internet of Things sensor network in real time, comprehensively cover the operation information of key components of the equipment, extract features from the multi-source monitoring data set, generate structure stability features and action fluency features, accurately reflect the mechanical properties of the equipment from different dimensions, and more deeply understand the operation state of the equipment. The structure stability features and the action fluency features are input into a pre-trained abnormality recognition model for abnormality evaluation, the generated equipment state evaluation result not only contains an abnormality existence identifier, but also explicitly indicates abnormality associated component information, greatly improves the accuracy and pertinence of abnormality recognition, extracts feature sequences in a historical period according to the abnormality associated component information and performs time sequence comparison analysis, accurately determines an abnormality influence range, avoids blind maintenance, and finally generates a supervision instruction containing a maintenance priority and a component identifier based on the comparison result of the abnormality influence range and the historical feature sequence, and sends the supervision instruction to a device management system to trigger a maintenance response, realizes accurate and efficient supervision of the mechanical parking equipment, effectively reduces the equipment maintenance cost, improves the reliability and safety of the equipment operation, and guarantees the normal operation of the parking lot. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 is a schematic diagram of the execution process of the parking equipment intelligent supervision method based on Internet of Things provided by the embodiment of the present application.

[0013] Figure 2 is a schematic diagram of exemplary hardware and software components of the parking equipment intelligent supervision system based on Internet of Things provided by the embodiment of the present application. DETAILED DESCRIPTION

[0014] The present application will be specifically described below in conjunction with the drawings of the specification, Figure 1 is a flowchart of the parking equipment intelligent supervision method based on Internet of Things provided by an embodiment of the present application, and the parking equipment intelligent supervision method based on Internet of Things will be described in detail below.

[0015] Step S110: Obtain a multi-source monitoring data set of the mechanical parking equipment collected by the Internet of Things sensor network in real time, the multi-source monitoring data set containing lifting mechanism displacement data, transverse mechanism power data and locking mechanism state data in a continuous time period.

[0016] In the scenario of intelligent parking management, in order to realize effective supervision of the mechanical parking equipment, first of all, the related data needs to be collected in real time by means of the Internet of Things sensor network. For the mechanical parking equipment, the coordinated work of multiple key mechanisms is involved in the running process.

[0017] In this embodiment, the Internet of Things sensor network can be deployed at various key positions of the mechanical parking equipment. At the lifting mechanism, a displacement sensor can be installed, which functions to monitor the displacement of the lifting mechanism in a continuous time period in real time. The displacement sensor can record the change of the position of the lifting mechanism with the movement of the lifting mechanism, and transmit the displacement data in the form of electrical signals. For example, when the lifting mechanism performs the actions of rising or falling, the displacement sensor can accurately capture the displacement value at each moment, and the displacement value constitutes the lifting mechanism displacement data.

[0018] As for the transverse mechanism, a power sensor can be installed. The transverse mechanism is responsible for the transverse movement of the vehicle in the parking equipment, and the stable output of the power is crucial for the normal operation of the equipment. The power sensor can monitor the power parameters of the transverse mechanism in real time, such as the output power and torque of the motor. Through the collection of these power data, the running state of the transverse mechanism under different working conditions can be understood. For example, during the transverse movement of the vehicle, the power sensor can continuously record the change of the power value, forming the transverse mechanism power data.

[0019] As for the locking mechanism, a state sensor can be installed. The main function of the locking mechanism is to lock the vehicle firmly after parking, to ensure the safety of the vehicle. The state sensor can detect the state of the locking mechanism in real time, to determine whether it is in the locked state or the unlocked state. Each time the state of the locking mechanism changes, the state sensor will record it and generate the locking mechanism state data.

[0020] The above data from different sensors will be transmitted to the data processing center through the network. In the transmission process, in order to ensure the accuracy and integrity of the data, encryption and error correction techniques can be used. For example, the data is encrypted using common encryption algorithms in related technologies to prevent data from being stolen or tampered with during transmission; at the same time, error correction code technology is used to detect and correct errors that may occur during transmission.

[0021] After receiving these multi-source monitoring data, the data processing center can preliminarily store and organize them. When storing, they can be classified and stored according to the type of data (lifting mechanism displacement data, horizontal movement mechanism power data, locking mechanism state data) and time stamp, so as to facilitate subsequent query and processing.

[0022] Step S120: Feature extraction is performed on the multi-source monitoring data set to generate structural stability features and action fluency features of the mechanical properties of the equipment, the structural stability features reflect the time consistency of the geometric shape of the key components, and the action fluency features reflect the continuity of the timing connection when performing actions.

[0023] After obtaining the multi-source monitoring data set, in order to deeply understand the running state of the mechanical parking equipment, feature extraction needs to be performed on the multi-source monitoring data set. Structural stability features and action fluency features are two important indicators that reflect the mechanical properties of the equipment from different angles.

[0024] Structural stability features mainly focus on the time consistency of the geometric shape of key components of the parking equipment. For example, the guide rails of the lifting mechanism, the transmission components of the horizontal movement mechanism, and other key components. If their geometric shape remains relatively stable during operation, it means that the structural performance of the equipment is good. If the geometric shape of the above components changes greatly within a continuous period of time, it may mean that the equipment has potential faults or safety hazards.

[0025] Action fluency features focus on the timing connection continuity of the equipment when performing actions. Each mechanism of the mechanical parking equipment needs to perform actions in a set order and time interval, such as the lifting mechanism rising to a specified position, then the horizontal movement mechanism moving horizontally, and finally the locking mechanism locking. If the timing connection between these actions is not continuous, it may lead to low equipment running efficiency, or even failure.

[0026] Step S121: Time stamp alignment processing is performed on the multi-source monitoring data set to obtain synchronized monitoring data.

[0027] Due to the differences in sampling frequency and start time of different sensors, the multi-source monitoring data collected may not be consistent in time. In order to ensure the accuracy of subsequent processing, time stamp alignment processing needs to be performed on the multi-source monitoring data set.

[0028] In this embodiment, the data processing center first checks the data collected by each sensor and extracts the time stamp information. Time stamp is an identifier that records the data collection time, through which the collection order and time point of the data can be determined.

[0029] Then, time interpolation and synchronization algorithms are used to process the data. For data with inconsistent timestamps, linear interpolation or other interpolation methods can be used to estimate the data value at the same time point based on the timestamps and values ​​of adjacent data. For example, if there is a difference in the timestamps of the lifting mechanism displacement data and the traversing mechanism power data, the overlapping area between them can be found, and then the displacement and power values ​​at the same time point can be calculated based on their respective data change trends.

[0030] Through the above timestamp alignment process, the final synchronized monitoring data is obtained. The displacement data of the lifting mechanism, the power data of the traversing mechanism, and the status data of the locking mechanism in the synchronized monitoring data are consistent in time.

[0031] Step S122: Perform time-series fluctuation analysis on the displacement data of the lifting mechanism in the synchronous monitoring data, and calculate the variance parameter of the displacement value in a continuous time period as the first component of the structural stability characteristics.

[0032] The displacement data of the lifting mechanism reflects its positional changes during operation. Time-series fluctuation analysis of this data can help understand the operational stability of the lifting mechanism.

[0033] Step S1221: Extract the displacement value sequence of the lifting mechanism in the continuous operation cycle from the synchronous monitoring data. The displacement value sequence includes the initial displacement value at the start time of each operation cycle and the final displacement value at the end time.

[0034] In synchronous monitoring data, the first step is to determine the operating cycle of the lifting mechanism. The operating cycle of the lifting mechanism refers to the time required for it to complete one full upward and downward movement. By analyzing the displacement data, the start and end times of each operating cycle can be identified.

[0035] For example, when the displacement data begins to show an upward trend, it can be considered the start of an action cycle; when the displacement data reaches its maximum value and then begins to decline, and falls back to near the initial position, it can be considered the end of that action cycle. Within each action cycle, the initial displacement value at the start time and the final displacement value at the end time are recorded. These initial and final displacement values ​​are arranged sequentially to form the displacement value sequence of the lifting mechanism within continuous action cycles.

[0036] Step S1222: Perform sliding window division processing on the displacement value sequence, where each sliding window contains a preset number of continuous action cycles.

[0037] To analyze the fluctuations in displacement data in greater detail, it is necessary to divide the displacement numerical sequence into sliding windows. A sliding window is a commonly used data processing method that moves across the data sequence, extracting a set length of data for analysis at each step.

[0038] In this embodiment, the preset number is a fixed value determined based on actual conditions. For example, if the preset number is set to 5, then each sliding window contains 5 consecutive action cycles. Starting from the first action cycle of the displacement value sequence, data from 5 action cycles are selected sequentially as the first sliding window; then the window is moved forward by one action cycle, and data from the next 5 action cycles are selected as the second sliding window, and so on, until the entire displacement value sequence has been traversed.

[0039] Step S1223: Calculate the average value of the initial displacement and the average value of the final displacement within each sliding window, and use them as the reference initial displacement and reference final displacement of the sliding window, respectively.

[0040] For the data within each sliding window, the average of its initial displacement value and the average of its final displacement value need to be calculated. Taking a sliding window with 5 motion cycles as an example, the initial displacement values ​​of these 5 motion cycles are added together and then divided by 5. The result is the reference initial displacement of the sliding window. Similarly, the final displacement values ​​of these 5 motion cycles are added together and then divided by 5. The result is the reference final displacement of the sliding window.

[0041] The aforementioned initial and final displacements can be used as reference values ​​for displacement data within the sliding window for subsequent fluctuation analysis.

[0042] Step S1224: Calculate the average of the squared differences between the initial displacement value and the reference initial displacement for each action cycle within the sliding window, and use this as the initial displacement fluctuation variance.

[0043] After obtaining the reference initial displacement for each sliding window, it is necessary to calculate the difference between the initial displacement value and the reference initial displacement for each motion cycle within the sliding window. For example, for the first motion cycle within the sliding window, the reference initial displacement is subtracted from its initial displacement value to obtain a difference. Then, this difference is squared to obtain the square of the difference.

[0044] The above calculations are performed for each motion cycle within the sliding window, resulting in a series of squared differences. These squared differences are summed and then divided by the number of motion cycles within the sliding window to obtain the initial displacement fluctuation variance. The initial displacement fluctuation variance reflects the degree of fluctuation of the initial displacement value within the sliding window relative to the reference initial displacement. A large initial displacement fluctuation variance indicates that the initial displacement value fluctuates drastically, and the lifting mechanism has poor stability at the starting position; conversely, a small initial displacement fluctuation variance indicates that the initial displacement value is relatively stable, and the lifting mechanism operates reliably at the starting position.

[0045] Step S1225: Calculate the average of the squared differences between the termination displacement value and the reference termination displacement for each action cycle within the sliding window, and use this as the variance of the termination displacement fluctuation.

[0046] Similar to calculating the initial displacement fluctuation variance, after obtaining the reference termination displacement for each sliding window, the difference between the termination displacement value for each action cycle within the sliding window and the reference termination displacement is calculated. These differences are squared, and then the squared differences are summed and divided by the number of action cycles within the sliding window to obtain the termination displacement fluctuation variance.

[0047] The variance of the termination displacement fluctuation reflects the degree of fluctuation of the termination displacement value relative to the reference termination displacement within the sliding window. It can be used to evaluate the stability of the lifting mechanism at the termination position. If the variance of the termination displacement fluctuation is large, it indicates that the lifting mechanism is less stable when reaching the designated position, which may affect the parking accuracy of the vehicle. Conversely, if the variance of the termination displacement fluctuation is small, it indicates that the lifting mechanism operates more stably at the termination position and can accurately park the vehicle at the designated position.

[0048] Step S1226: The sum of the initial displacement fluctuation variance and the final displacement fluctuation variance is used as the displacement value variance parameter of the sliding window. The displacement value variance parameter is used to characterize the position stability of the lifting mechanism in a continuous action cycle.

[0049] Finally, the initial displacement fluctuation variance and the final displacement fluctuation variance are added together, and the sum is the displacement variance parameter of the sliding window. This displacement variance parameter comprehensively reflects the stability of the lifting mechanism at the starting and ending positions within a continuous operating cycle.

[0050] If the displacement variance parameter is large, it indicates that the position of the lifting mechanism fluctuates significantly within a continuous operating cycle, resulting in poor structural stability and potential issues such as guide rail deformation and wear of transmission components. Conversely, if the displacement variance parameter is small, it indicates that the lifting mechanism's position is relatively stable and its structural performance is good. This displacement variance parameter serves as the first component of the structural stability characteristic and is used for subsequent anomaly assessment.

[0051] Step S123: Perform frequency component decomposition processing on the lateral movement mechanism power data in the synchronous monitoring data, and extract the frequency concentration parameter of the power value within the action cycle as the first component of the action smoothness feature.

[0052] The power data of the traverse mechanism reflects its power output during operation. By performing frequency component decomposition on the power data, the frequency characteristics of the traverse mechanism's power output can be understood, thereby assessing the smoothness of its operation.

[0053] Step S1231: Extract the power value sequence of the transverse mechanism within a single complete action cycle from the synchronous monitoring data. The power value sequence includes the power values ​​of the action start-up phase, the action execution phase, and the action stop phase.

[0054] In the synchronous monitoring data, the first step is to determine the single complete motion cycle of the traverse mechanism. The single complete motion cycle of the traverse mechanism includes the motion initiation phase, the motion execution phase, and the motion stopping phase. During the motion initiation phase, the traverse mechanism needs to overcome the inertia of its stationary state and output a large amount of power to initiate the movement; during the motion execution phase, the traverse mechanism maintains a stable power output to achieve the lateral movement of the vehicle; during the motion stopping phase, the traverse mechanism needs to gradually reduce the power output to bring the vehicle to a smooth stop.

[0055] The power value sequence within a complete action cycle is extracted from the synchronous monitoring data, and the power value at each stage is recorded. For example, during the action start-up stage, the process of the power value rising from 0 to a stable value is recorded; during the action execution stage, the stable power output value is recorded; and during the action stop stage, the process of the power value decreasing from a stable value to 0 is recorded. These power values ​​constitute the power value sequence of the traverse mechanism within a single complete action cycle.

[0056] Step S1232: Perform a fast Fourier transform on the dynamic numerical sequence to generate a frequency distribution map of the dynamic values, wherein the frequency distribution map contains the energy intensity corresponding to different frequency components.

[0057] To analyze the frequency characteristics of dynamic numerical sequences, it is necessary to perform Fast Fourier Transform (FFT) processing on them. FFT is a mathematical method for converting time-domain signals into frequency-domain signals. It can be used to transform dynamic numerical sequences from the time domain to the frequency domain, obtaining their frequency distribution.

[0058] In this embodiment, the extracted dynamic numerical sequence is input into a Fast Fourier Transform (FFT) algorithm. The FFT algorithm performs a series of mathematical operations on the dynamic numerical sequence, decomposing it into combinations of sine and cosine waves with different frequency components. Each frequency component corresponds to an energy intensity, and these energy intensities constitute the frequency distribution map of the dynamic values.

[0059] For example, in a frequency distribution plot, the horizontal axis represents frequency, and the vertical axis represents energy intensity. Different frequency components correspond to different energy peaks. By observing the frequency distribution plot, one can understand the distribution of each frequency component in the dynamic numerical sequence.

[0060] Step S1233: Identify the dominant frequency component with the highest energy intensity and its corresponding energy percentage in the frequency distribution diagram.

[0061] After obtaining the frequency distribution diagram of the power values, it is necessary to identify the dominant frequency component with the highest energy intensity. The dominant frequency component reflects the main frequency characteristics of the power output of the traverse mechanism.

[0062] By traversing the frequency distribution map, the frequency component with the highest energy intensity is identified and determined as the dominant frequency component. Then, the energy proportion corresponding to the dominant frequency component is calculated, which is the ratio of the energy intensity of the dominant frequency component to the sum of the energy intensities of all frequency components.

[0063] For example, if the energy intensity of the dominant frequency component is E1, and the sum of the energy intensities of all frequency components is E_total, then the energy proportion of the dominant frequency component is E1 / E_total. This energy proportion reflects the importance of the dominant frequency component in the entire frequency distribution.

[0064] Step S1234: Calculate the ratio of the energy proportion of the main frequency component to the energy proportion of the secondary frequency component, and use it as the frequency concentration parameter of the dynamic value.

[0065] After identifying the dominant frequency component, it is also necessary to find the secondary frequency component. The secondary frequency component refers to the frequency component with relatively high energy intensity, other than the dominant frequency component.

[0066] Calculate the energy proportion of the secondary frequency component, which is the ratio of the energy intensity of the secondary frequency component to the sum of the energy intensities of all frequency components. Then divide the energy proportion of the primary frequency component by the energy proportion of the secondary frequency component; the resulting ratio is the frequency concentration parameter of the dynamic value.

[0067] If the frequency concentration parameter is large, it means that the main frequency component dominates the power output, the frequency of the power output is relatively concentrated, and the movement of the traverse mechanism is relatively smooth; conversely, if the frequency concentration parameter is small, it means that the frequency of the power output is relatively dispersed, and the movement of the traverse mechanism may be stuck or unstable.

[0068] Step S1235: Perform exponential smoothing on the frequency concentration parameters of multiple consecutive action cycles to generate a smooth concentration parameter that reflects the stability of the lateral movement frequency. The smooth concentration parameter is used to characterize the smoothness of the lateral movement mechanism's action execution.

[0069] To more accurately assess the smoothness of the traverse mechanism's movements, it is necessary to process the frequency concentration parameter across multiple consecutive movement cycles. Exponential smoothing is a commonly used method for smoothing time series data. It applies a weighted average to the data, giving more weight to recent data and thus better reflecting the data's changing trends.

[0070] In this embodiment, the frequency concentration parameters of multiple consecutive action cycles are input into the exponential smoothing algorithm. The exponential smoothing algorithm can perform weighted calculations on each frequency concentration parameter according to a preset smoothing coefficient. For example, for the frequency concentration parameter Cn of the nth action cycle, the exponential smoothing algorithm will calculate the current smooth concentration parameter Sn based on the previous smoothed concentration parameter S(n-1) and the current frequency concentration parameter Cn. The calculation formula is Sn=α*Cn+(1-α)*S(n-1), where α is the smoothing coefficient, and its value ranges from 0 to 1.

[0071] By exponentially smoothing the frequency concentration parameter across multiple consecutive motion cycles, the resulting smoothed concentration parameter better reflects the frequency stability of the traverse mechanism's motion execution. A stable smoothed concentration parameter indicates a concentrated power output frequency and smooth motion execution; conversely, large fluctuations in the smoothed concentration parameter suggest potential instability in the traverse mechanism's motion. This smoothed concentration parameter serves as the first component of the motion smoothness characteristic for subsequent anomaly assessment.

[0072] Step S124: Perform statistical processing on the state duration of the locking mechanism state data in the synchronous monitoring data, and calculate the standard deviation parameter of the locking state holding time as the second component of the structural stability feature.

[0073] The state data of the locking mechanism reflects its locking and unlocking status during operation. By statistically processing the duration of the locked state, the stability of the locking mechanism can be evaluated, and thus used as part of the structural stability characteristics.

[0074] Step S1241: Extract the state change information of the locking mechanism from the synchronous monitoring data, and determine the start and end times of the locking state.

[0075] In the synchronous monitoring data, the locking mechanism status data exists in the form of status values, such as 0 representing the unlocked state and 1 representing the locked state. By analyzing these status values, the time point when the status value changes from 0 to 1 is the start time of the locking state; the time point when the status value changes from 1 to 0 is the end time of the locking state.

[0076] For example, when the state sensor records that the state value changes from 0 to 1, the time t1 is recorded as the start time of the locked state; when the state value changes from 1 to 0, the time t2 is recorded as the end time of the locked state.

[0077] Step S1242: Calculate the duration of each locking state, i.e., the difference between the end time and the start time.

[0078] After determining the start and end times of the locking state, the duration of each locking state is obtained by subtracting the start time from the end time. For example, for a locking state with a start time of t1 and an end time of t2, the duration of the locking state is T = t2 - t1.

[0079] Step S1243: Count the duration of holding multiple consecutive locking states and calculate the standard deviation parameter of these durations.

[0080] The duration of holding multiple consecutive lockout states is statistically analyzed to obtain a set of duration data. Then, the standard deviation of this duration data is calculated. Standard deviation is a statistic that measures the dispersion of data, reflecting the fluctuation of the data relative to the mean.

[0081] First, calculate the average of this set of duration data. Add up the durations of all locked states, then divide by the number of locked states to obtain the average. Next, for each locked state duration, calculate the difference between it and the average, square the difference, add up the squares of all differences, and divide by the number of locked states. Finally, process the result (not explicitly expressed as a square root, but essentially meaning the same thing) to obtain the standard deviation. The standard deviation reflects the fluctuation of the locked state duration. A large standard deviation indicates unstable locked state durations, suggesting a possible malfunction in the locking mechanism, such as wear on the locking tongue or abnormalities in the control system, which would affect the overall structural stability of the equipment. Conversely, a small standard deviation indicates relatively stable locked state durations, suggesting reliable operation of the locking mechanism and contributing to the maintenance of the equipment's structural stability. This standard deviation serves as the second component of the structural stability characteristic.

[0082] Step S125: Perform correlation coupling analysis on the power data of the lateral movement mechanism and the displacement data of the lifting mechanism in the synchronous monitoring data, and extract the correlation coefficient parameter between the power value and the displacement value as the second component of the motion smoothness feature.

[0083] There may be a certain correlation between the power data of the traverse mechanism and the displacement data of the lifting mechanism. Correlation and coupling analysis can reveal this relationship and thus assess the smoothness of the equipment's operation.

[0084] Step S1251: Simultaneously extract the power data sequence of the traverse mechanism and the displacement data sequence of the lifting mechanism from the synchronous monitoring data to ensure that the two sequences correspond in time.

[0085] In the synchronous monitoring data, the power data of the traversing mechanism and the displacement data of the lifting mechanism are extracted in chronological order. Since timestamp alignment has been performed previously, these two data sequences are one-to-one in time. For example, at the same time t, the corresponding power value of the traversing mechanism is P(t), and the displacement value of the lifting mechanism is D(t). This results in two time-corresponding sequences, denoted as the power data sequence {P(t1), P(t2), ...} and the displacement data sequence {D(t1), D(t2), ...}.

[0086] Step S1252: Perform some kind of standardization on the dynamic data sequence and displacement data sequence to make them have the same dimensions and data range.

[0087] Because dynamic data and displacement data have different dimensions, they need to be standardized to perform effective correlation analysis. The purpose of standardization is to transform data with different dimensions to a unified scale. Common standardization methods can be used, such as subtracting the mean of the sequence from each data point and then dividing by the standard deviation of the sequence. For a dynamic data sequence {P(t)}, let its mean be μP and its standard deviation be σP, then the standardized dynamic data is (P(t) - μP) / σP; for a displacement data sequence {D(t)}, let its mean be μD and its standard deviation be σD, then the standardized displacement data is (D(t) - μD) / σD. After this processing, the data range and dimensions of the two sequences are unified, facilitating subsequent calculations.

[0088] Step S1253: Calculate the correlation coefficient between the standardized dynamic data sequence and the displacement data sequence.

[0089] The correlation coefficient is an indicator that measures the degree of linear correlation between two variables. In this embodiment, the correlation coefficient between the standardized power data sequence and the displacement data sequence is calculated. For example, a common correlation coefficient calculation method, such as the Pearson correlation coefficient method, can be used. The calculation process is roughly as follows: first, calculate the sum of the products of corresponding data points in the two sequences, and then divide by a comprehensive measure of the sum of squares of the two sequences. If the correlation coefficient is close to 1, it indicates a strong positive linear correlation between the power data and the displacement data, meaning that changes in power are closely accompanied by corresponding changes in displacement. This indicates that the movement coordination between the traversing mechanism and the lifting mechanism is very good, and the equipment movement is relatively smooth. If the correlation coefficient is close to -1, it indicates a strong negative linear correlation. If the correlation coefficient is close to 0, it indicates a weak linear relationship between the two, which may indicate a problem with the movement coordination between the traversing mechanism and the lifting mechanism, and the smoothness of the equipment movement may be poor. This correlation coefficient serves as the second component of the smoothness characteristic of the movement.

[0090] Step S126: Perform weighted fusion processing on the first and second components of the structural stability feature and the first and second components of the motion smoothness feature to generate the final structural stability feature and motion smoothness feature, respectively.

[0091] In order to make full use of the various feature components extracted above, a weighted fusion process is required.

[0092] Step S1261: Determine the weights of the first and second components of the structural stability feature.

[0093] The weights of different components are determined based on their relative importance in reflecting structural stability. For example, the first component of the structural stability characteristic (the variance parameter of the displacement value of the lifting mechanism) mainly reflects the positional stability of the lifting mechanism, while the second component (the standard deviation parameter of the duration of the locking state of the locking mechanism) reflects the stability of the locking mechanism. If the positional stability of the lifting mechanism is considered more critical to the overall structural stability, then a larger weight can be assigned to the first component and a smaller weight to the second component. Let the weight of the first component be w1, the weight of the second component be w2, and w1 + w2 = 1.

[0094] Step S1262: Perform weighted fusion of the first and second components of the structural stability characteristics.

[0095] The first component of the structural stability feature is multiplied by its weight w1, and the second component is multiplied by its weight w2. The two results are then concatenated. For example, if the first component is represented by S1 and the second component by S2, then the fused structural stability feature S is composed of w1*S1 and w2*S2. This structural stability feature comprehensively considers the stability of both the lifting and locking mechanisms, providing a more complete reflection of the equipment's structural stability.

[0096] Step S1263: Determine the weights of the first and second components of the motion smoothness feature.

[0097] Similarly, the weights of the first component (frequency concentration parameter of the dynamic value) and the second component (correlation coefficient parameter between the dynamic value and the displacement value) of the motion smoothness feature are determined based on their relative importance in reflecting motion smoothness. If the frequency concentration of the dynamic output is considered to be more important to motion smoothness, then a larger weight can be assigned to the first component, denoted as w3; and a smaller weight can be assigned to the second component, denoted as w4, with w3+w4=1.

[0098] Step S1264: Perform weighted fusion of the first and second components of the motion smoothness feature.

[0099] The first component of the motion smoothness feature is multiplied by its weight w3, and the second component is multiplied by its weight w4. The two results are then concatenated. For example, if the first component is represented by F1 and the second component by F2, then the fused motion smoothness feature F is composed of w3*F1 and w4*F2. This method of obtaining the motion smoothness feature comprehensively considers the power output frequency and the relationship between power and displacement, thus more accurately reflecting the motion smoothness of the equipment.

[0100] Step S130: Input the structural stability features and the motion smoothness features into the pre-trained anomaly recognition model for anomaly evaluation, and generate a device status evaluation result containing anomaly presence identifiers and information on anomaly-related components.

[0101] After obtaining the structural stability features and motion smoothness features, a pre-trained anomaly detection model is needed to assess whether the device has any anomalies.

[0102] Step S131: Input the structural stability features and the motion smoothness features into the input layer of the anomaly recognition model, and perform standardized feature vector mapping processing on the structural stability features and the motion smoothness features to generate standardized feature vectors.

[0103] The input layer of the anomaly detection model first receives structural stability features and motion fluency features. Since the raw data for these two features may have different scales and ranges, a standardized feature vector mapping process is needed to enable the model to better process this data. Standardized feature vector mapping transforms the structural stability features and motion fluency features into standardized feature vectors with a uniform scale and range. A similar standardization method can be used, subtracting the mean of the feature sequence from each feature value and then dividing by the standard deviation of the feature sequence. After this processing, the resulting standardized feature vector contains comprehensive information on structural stability and motion fluency, exhibiting better comparability and processability, and can be more effectively input into subsequent layers of the anomaly detection model for processing.

[0104] Step S132: Perform pairwise correlation analysis on the standardized feature vectors through the feature correlation layer of the anomaly recognition model to generate a correlation feature matrix that reflects the interaction between structural stability and motion smoothness.

[0105] After receiving the standardized feature vectors, the feature association layer of the anomaly detection model performs pairwise correlation analysis. This layer analyzes the relationships between structural stability features and motion smoothness features. For example, it analyzes how changes in structural stability affect motion smoothness, or whether anomalies in motion smoothness are related to structural stability. Through this pairwise correlation analysis, a correlation feature matrix is ​​generated. Each element in the correlation feature matrix represents the degree of correlation between different dimensions of the structural stability and motion smoothness features. This correlation feature matrix can reveal the interaction between device structural stability and motion smoothness in greater depth.

[0106] Step S133: Use the temporal memory layer of the anomaly recognition model to perform historical data matching processing on the associated feature matrix, and extract the similarity measure value between the current associated feature and the historical normal associated feature.

[0107] The anomaly detection model's temporal memory layer stores historical normal correlation feature sequences, which are collected and processed during normal device operation. Upon receiving the current correlation feature matrix, the temporal memory layer matches it with the historical normal correlation feature sequences.

[0108] Step S1331: Extract historical normal correlation feature sequences that are consistent with the current time window length from the historical database of the anomaly identification model.

[0109] First, historical normal association feature sequences with the same time window length as the current one are selected from the historical database of the anomaly detection model. For example, if the association feature matrix being processed was generated within a one-hour time window, then historical normal association feature sequences within the same one-hour time window are extracted from the historical database. This ensures that the two sequences being matched are consistent in time scale, improving matching accuracy.

[0110] Step S1332: The current associated feature matrix and the historical normal associated feature sequence are input into the dynamic time warping algorithm for sequence alignment processing to generate an aligned feature matching path.

[0111] Dynamic time warping (VTW) is an algorithm for processing time series matching. By inputting the current correlation feature matrix and historical normal correlation feature sequences into VTW, the current correlation feature matrix and historical normal correlation feature sequences can be aligned. Since real-world time series may exhibit temporal scaling and distortion, VTW can find the optimal matching path to better align the two sequences in time. For example, it can adjust the matching relationship of corresponding elements in the two sequences to minimize their differences. After processing, an aligned feature matching path is generated, which records the matching status between the current correlation feature matrix and the historical normal correlation feature sequences.

[0112] Step S1333: Calculate the sum of Euclidean distances between corresponding feature values ​​on the matching path, and use it as the difference parameter between the current associated feature and the historical normal associated feature.

[0113] After obtaining the aligned feature matching path, the Euclidean distance between corresponding feature values ​​on the matching path is calculated. Euclidean distance is a common method for measuring the distance between two vectors. For each pair of corresponding feature values ​​on the matching path, their Euclidean distance is calculated, and then all these Euclidean distances are summed. The sum is the difference parameter between the current associated feature and the historical normal associated features. This difference parameter reflects the degree of difference between the current associated feature and the historical normal associated features. If the difference parameter is large, it indicates that the current associated feature deviates significantly from the historical normal situation, and the device may be malfunctioning; conversely, if the difference parameter is small, it indicates that the current situation is relatively similar to the historical normal situation, and the device may be in normal operating condition.

[0114] Step S1334: Input the difference parameter into the similarity conversion function to generate a similarity metric value. The larger the similarity metric value, the more similar the current associated feature is to the historical normal associated feature.

[0115] To more intuitively represent the similarity between current associated features and historical normal associated features, the difference parameter is input into the similarity transformation function. The similarity transformation function converts the difference parameter into a similarity measure. This transformation process reverses the difference, so that the greater the difference, the smaller the similarity measure; and vice versa. For example, a simple linear transformation can be used to apply a linear transformation to the difference parameter to obtain the similarity measure. This similarity measure can more directly reflect the similarity between the current associated feature and historical normal associated features, and is used for subsequent anomaly detection.

[0116] Step S134: The similarity metric is processed by the decision output layer of the anomaly identification model to determine a threshold. If the similarity metric is lower than the preset threshold, an anomaly presence flag is output. The responsibility attribution module of the anomaly identification model is used to locate the feature component that contributes the most to the decrease in the similarity metric and determine the corresponding anomaly associated component information. If the similarity metric is higher than or equal to the preset threshold, a normal presence flag is output.

[0117] After receiving the similarity metric, the decision output layer of the anomaly detection model compares it with a preset threshold. This preset threshold is a critical value determined during model training based on a large amount of normal and abnormal data. If the similarity metric is lower than the preset threshold, it indicates a significant difference between the current associated features and historical normal associated features, suggesting a possible equipment anomaly. At this point, the anomaly detection model's responsibility attribution module is activated. This module analyzes the contribution of each feature component to the decrease in the similarity metric. Using conventional analysis methods (such as sensitivity analysis), it identifies the feature component that contributes the most to the decrease in the similarity metric. For example, if a certain dimension of the structural stability features is found to contribute the most to the decrease in the similarity metric, then the abnormal associated component can be identified as a component related to structural stability, such as a lifting mechanism or locking mechanism. Based on these analysis results, the corresponding abnormal associated component information is determined, and an anomaly presence indicator is output. This abnormal associated component information is also included in the equipment status assessment results. If the similarity metric is higher than or equal to the preset threshold, it indicates that the current associated features are similar to historical normal associated features, and the equipment is in normal operating condition. The decision output layer then outputs a normal presence indicator.

[0118] Step S140: Based on the abnormal associated component information in the equipment status assessment results, extract the feature sequences of the corresponding abnormal associated components in the historical time period and perform time-series comparison analysis to determine the scope of abnormal impact in the mechanical parking equipment.

[0119] After obtaining the equipment status assessment results, if they include information about abnormal and related components, it is necessary to further analyze the scope of the abnormality's impact.

[0120] Step S141: Based on the abnormal associated component information, extract the feature sequence of the abnormal associated component within a preset time period from the historical storage records of the multi-source monitoring data set. The feature sequence includes structural stability feature components and motion smoothness feature components.

[0121] Based on the information of abnormally associated components in the equipment condition assessment results, the feature sequence of the abnormally associated component over a preset period of time is extracted from the historical storage records of the multi-source monitoring data set. For example, if the abnormally associated component is a lifting mechanism, then the structural stability feature components (such as displacement variance parameters) and motion smoothness feature components (such as the correlation coefficient with the power data of the traverse mechanism) of the lifting mechanism over a period of time are extracted from the historical storage records. The above feature sequence records the changes in the operating status of the abnormally associated component over a period of time.

[0122] Step S142: Perform sliding window division processing on the historical feature sequence, with each sliding window containing the same number of consecutive time points.

[0123] To facilitate the analysis of historical feature sequences' changing trends, a sliding window partitioning process is applied. Sliding window partitioning divides the historical feature sequence into segments of a predetermined length. Each sliding window contains the same number of consecutive time points. For example, the historical feature sequence can be divided into windows of 10 consecutive time points. Starting from the first time point of the historical feature sequence, 10 consecutive time points are selected as the first sliding window; then the window is moved forward by one time point, and the next 10 time points are selected as the second sliding window, and so on, until the entire historical feature sequence has been traversed. Through this sliding window partitioning process, the historical feature sequence can be divided into multiple subsequences, facilitating the individual analysis of each subsequence.

[0124] Step S143: Calculate the average value of the feature values ​​within each sliding window as the baseline feature value of that sliding window.

[0125] For each feature value within a sliding window, its average value is calculated and used as the baseline feature value for that sliding window. For example, for a sliding window containing 10 time points, the feature values ​​at these 10 time points are added together and then divided by 10; the result is the baseline feature value for that sliding window. The baseline feature value represents the average level of the features within the sliding window and is used for subsequent comparisons with the feature values ​​of the current time period.

[0126] Step S144: Calculate the absolute difference between the feature value of the abnormal associated component and the baseline feature value of each sliding window in the current time period, and identify the sliding window with the largest absolute difference as the abnormal starting associated window.

[0127] After obtaining the baseline feature value for each sliding window, the absolute difference between the feature value of the anomaly-related component in the current time period and the baseline feature value of each sliding window is calculated. For example, if the feature value of the anomaly-related component in the current time period is C, and the baseline feature value of a certain sliding window is B, then their absolute difference is |CB|. This calculation is performed on the baseline feature values ​​of all sliding windows, resulting in a series of absolute differences. Then, the sliding window with the largest absolute difference is identified and used as the anomaly initiation window. The time range corresponding to the anomaly initiation window may be the time range in which the anomaly first appeared; by determining this time range, the origin of the anomaly can be traced more accurately.

[0128] Step S145: Extract the time range corresponding to the abnormal start-up association window, and combine it with the correlation analysis of the equipment's mechanical structure to determine the changes in the characteristics of other components affected by the abnormal component within this time range.

[0129] After identifying the anomaly initiation correlation window, its corresponding time range is extracted. Then, combined with the correlation analysis of the equipment's mechanical structure, the characteristic changes of other components affected by the abnormal component within that time range are determined. For example, if the abnormally associated component is a lifting mechanism, based on the equipment's mechanical structure, an anomaly in the lifting mechanism may affect the traversing mechanism and the locking mechanism. By analyzing the characteristic changes of the traversing mechanism and the locking mechanism within the time range corresponding to the anomaly initiation correlation window, such as changes in the traversing mechanism's power data and changes in the locking mechanism's state, the degree of influence of the abnormal component on other components can be understood. By comparing the characteristic value changes of these components before and after the anomaly initiation correlation window, it can be determined whether they have been affected by the anomaly.

[0130] Step S146: Based on the changes in the characteristics of the other components, generate a description of the abnormal impact range, including the abnormal components and the affected components.

[0131] Step S1461: Perform threshold judgment processing on the feature change values ​​of the other components, and identify the components whose feature change values ​​exceed the preset influence threshold as directly affected components.

[0132] Based on the characteristic changes of other components, a threshold judgment is performed on the characteristic change value of each component. The preset impact threshold is a critical value determined based on the normal operating range of the equipment and experience. If the characteristic change value of a component exceeds the preset impact threshold, it indicates that the component is directly affected by the abnormal component, and it is identified as a directly affected component. For example, if the change value of the power data of the traverse mechanism within the time range corresponding to the abnormal start-up correlation window exceeds the preset impact threshold, then the traverse mechanism is identified as a directly affected component.

[0133] Step S1462: Perform a secondary threshold judgment process on the feature change value of the directly affected component to identify the component whose feature change is caused by the state change of the directly affected component as the indirectly affected component.

[0134] For directly affected components, a secondary threshold judgment is performed on their characteristic change values. This secondary threshold judgment is to further identify indirectly affected components that are affected by changes in the state of the directly affected components. A preset secondary impact threshold, determined based on equipment operating characteristics and experience, is used to determine whether a component's characteristic change is caused by a change in the state of the directly affected component. When the characteristics of a directly affected component change, this change may be transmitted to other components through the equipment's mechanical structure or operating logic. For each potentially affected component, its characteristic change value is calculated and compared with the secondary impact threshold. If a component's characteristic change value exceeds the secondary impact threshold, the component is considered to have its characteristic change caused by a change in the state of the directly affected component and is identified as an indirectly affected component. For example, when the power data of a traverse mechanism (a directly affected component) changes significantly, it may cause a change in the characteristics of an auxiliary support component that is mechanically connected or operationally related to it. If the characteristic change value of this auxiliary support component exceeds the secondary impact threshold, then it is identified as an indirectly affected component.

[0135] Step S1463: Count the number and distribution of the abnormal components, directly affected components and indirectly affected components.

[0136] The number of abnormal, directly affected, and indirectly affected components needs to be counted, requiring a one-by-one count of each identified component. Simultaneously, the distribution of these components within the mechanical parking system must be determined. In the equipment's design and layout, each component has a specific installation location. This location can be clearly identified through design drawings, installation records, or sensor location information. For example, the lifting mechanism for abnormal components might be located in the vertical lifting channel, the directly affected lateral movement mechanism in the horizontal movement track, and the indirectly affected auxiliary support components at the connection point between the lateral movement mechanism and the main structure.

[0137] Step S1464: Based on the quantity and distribution location, generate an anomaly impact range description that includes a list of component identifiers and a spatial distribution diagram.

[0138] The component identification list contains unique identifiers for abnormal components, directly affected components, and indirectly affected components. These identifiers can be serial numbers assigned to each component during equipment manufacturing or specific codes assigned to each component in the equipment management system. These identifiers allow for accurate identification and differentiation of different components. The spatial distribution diagram reflects the relative positions of each component within the parking equipment. Based on the equipment's 3D model or 2D layout diagram, the locations of abnormal, directly affected, and indirectly affected components can be marked on the diagram, visually demonstrating their spatial relationships. This allows maintenance personnel to quickly locate specific components using the component identification list and understand the relative positions and impact ranges of components using the spatial distribution diagram, providing clear guidance for subsequent maintenance work.

[0139] Step S150: Based on the comparison results of the abnormal impact range and historical feature sequence, generate a regulatory instruction containing maintenance priority and component identification, and send the regulatory instruction to the equipment management system to trigger a maintenance response.

[0140] After determining the scope of the anomaly and completing the comparison of historical feature sequences, corresponding regulatory instructions need to be generated to ensure timely maintenance of the equipment.

[0141] Step S151: Perform fault impact assessment on the abnormal components, directly affected components, and indirectly affected components within the scope of the abnormality. The fault impact assessment is based on the functional importance of each component during equipment operation and the magnitude of characteristic changes.

[0142] The impact of the anomaly on all components within its affected area, including the anomalous component, directly affected components, and indirectly affected components, is assessed. The assessment is based on two main aspects: the functional importance of each component in equipment operation and the magnitude of its characteristic changes. Each component performs different functions in equipment operation. Some components are core components, such as lifting mechanisms and traversing mechanisms, whose normal operation directly affects the basic functionality of the equipment; while others are auxiliary components with relatively minor functions. Functional importance can be assessed and graded based on equipment design documents and operating specifications. The magnitude of characteristic changes reflects the degree of deviation between the component's current state and its normal state; the larger the characteristic change, the more severe the anomaly. For example, for the anomalous lifting mechanism, if its displacement data changes significantly, and the lifting mechanism is a key component for vertical vehicle transport, its impact from the anomaly will be relatively high. Conversely, for an auxiliary support component, even if its characteristics change, its impact may be relatively low due to its relatively minor function. By comprehensively considering these two factors, the impact of each component's failure is assessed.

[0143] Step S152: Based on the assessment results of the degree of impact of the fault, sort the maintenance priorities of each component, and give higher priority to components with high functional importance and large characteristic change values.

[0144] Based on the assessment of the impact of the fault, the components within the affected area are prioritized for maintenance. Considering both functional importance and the magnitude of characteristic changes, components with high functional importance and large characteristic changes are assigned higher maintenance priority. For example, the lifting mechanism, a component with high functional importance and large characteristic changes, is prioritized; while indirectly affected components with relatively minor functional changes and smaller characteristic changes are prioritized. This prioritization method ensures that maintenance resources are used to address faults in components that have the greatest impact on equipment operation, improving maintenance efficiency and equipment recovery speed.

[0145] Step S153: Extract the component identifier and corresponding priority level from the maintenance priority sorting.

[0146] Extract the identifier and corresponding priority level of each component from the maintenance priority ranking results. The component identifier is unique to each component, such as the number or code mentioned earlier; the priority level is determined by the ranking and clarifies the order in which the component is maintained. For example, the lifting mechanism with component identifier "L001" has a priority level of "Level 1," indicating that it requires priority maintenance; the auxiliary support component with component identifier "A003" has a priority level of "Level 3," indicating that its maintenance order is relatively later. Organizing these component identifiers and priority levels provides specific information for generating subsequent regulatory instructions.

[0147] Step S154: Based on the comparison results of the historical feature sequences, generate maintenance guidance information that includes component maintenance order and maintenance content suggestions.

[0148] For example, step S154 may include:

[0149] Step S1541: Extract the abnormal start time from the historical feature sequence alignment results and calculate the abnormal duration from the abnormal start time to the current time.

[0150] The anomaly start time is identified from the comparison results of historical feature sequences. The anomaly start time can be determined by the time range corresponding to the previously defined anomaly start association window, marking the moment when the component begins to exhibit anomalies. The duration from the anomaly start time to the current time is calculated to obtain the anomaly duration. This anomaly duration reflects the duration of the component's abnormal state. For example, if the anomaly duration of a component is relatively long, it may indicate a more severe degree of damage, requiring more comprehensive maintenance measures.

[0151] Step S1542: Perform correlation analysis on the duration of the anomaly and the trend of feature changes to identify the correlation between the rate of feature change and the duration.

[0152] A correlation analysis is performed between the duration of the anomaly and the trend of characteristic changes. The trend of characteristic changes can be determined by observing the changes in component characteristic values ​​over time in historical characteristic sequences, such as whether the characteristic value shows an upward trend, a downward trend, or a fluctuating trend. Simultaneously, the rate of characteristic change is calculated, i.e., the amount of change in the characteristic value per unit time. Through correlation analysis, the correlation between the rate of characteristic change and the duration of the anomaly is identified. If the rate of characteristic change is fast and the duration of the anomaly is long, it indicates that the component's anomaly is worsening and may require more urgent and thorough maintenance; if the rate of characteristic change is slow and the duration of the anomaly is short, the component's anomaly is relatively mild. For example, for the power data of a traverse mechanism, if its characteristic change rate is fast and the anomaly has persisted for a long time, then it requires close attention and more effective maintenance measures.

[0153] Step S1543: For components whose feature change rate is greater than the first set change rate and whose duration is greater than the first set duration, generate maintenance recommendations with component replacement as the content.

[0154] The first set change rate and first set duration are determined based on the equipment's performance characteristics and experience. When the characteristic change rate of a component exceeds the first set change rate and the abnormality lasts for more than the first set duration, it indicates that the component's damage is already quite severe. Continued use may have a greater impact on the overall operation of the equipment. In this case, a maintenance recommendation based on component replacement is generated. For example, if the displacement data change rate of the lifting mechanism exceeds the first set change rate and the abnormality has lasted for more than the first set duration, it is recommended to directly replace the lifting mechanism to ensure the normal operation of the equipment.

[0155] Step S1544: For components whose feature change rate is less than the second set change rate and whose duration is less than the second set duration, generate maintenance suggestions with component debugging as the content, wherein the first set change rate is less than the second set change rate and the first set duration is less than the second set duration.

[0156] A second set change rate and a second set duration are set, where the first set change rate is less than the second set change rate, and the first set duration is less than the second set duration. When the characteristic change rate of a component is less than the second set change rate and the duration of the anomaly is less than the second set duration, it indicates that the component's anomaly is relatively minor and may only be caused by some minor adjustment issues. In this case, maintenance suggestions focusing on component adjustments are generated. For example, for a certain auxiliary sensor component with a small characteristic change rate and a short duration of anomaly, adjustments such as parameter calibration and connection checks can be suggested to restore its normal operation.

[0157] Step S1545: Based on the urgency of the maintenance recommendations, adjust the order of components in the maintenance priority ranking, prioritizing components that need to be replaced over components that need to be debugged.

[0158] Based on the urgency of maintenance recommendations, the order of components in the previous maintenance priority ranking is adjusted. Components requiring replacement typically indicate more severe faults and a greater impact on equipment operation, thus requiring priority; while components requiring debugging have relatively minor faults and can be addressed later. Replacement components are placed earlier in the maintenance sequence, and debugging components later. For example, in the original maintenance priority ranking, an auxiliary component requiring debugging was earlier, while a core component requiring replacement was later. After the adjustment, the core component requiring replacement is moved to the front, ensuring that maintenance resources are prioritized for handling more urgent faults.

[0159] Step S1546: Based on the adjusted maintenance sequence and corresponding maintenance recommendations, generate maintenance guidance information that includes component identification, maintenance type and execution sequence.

[0160] Based on the adjusted maintenance sequence and corresponding maintenance recommendations, detailed maintenance guidance information is generated. This guidance includes component identification, maintenance type (component replacement or component debugging), and execution order. The component identification accurately locates the component requiring maintenance; the maintenance type specifies the detailed maintenance content; and the execution order guides maintenance personnel to perform maintenance operations in the correct sequence. For example, the guidance might show: component identification "L001," maintenance type: component replacement, execution order: number 1; component identification "A003," maintenance type: component debugging, execution order: number 3. This allows maintenance personnel to efficiently perform maintenance work based on the guidance information.

[0161] Step S155: The maintenance priority sorting, component identification and maintenance guidance information are formatted and encapsulated to generate regulatory instructions that conform to the interface specifications of the equipment management system.

[0162] The maintenance priority ranking, component identification, and maintenance guidance information are encapsulated in a specific format. The equipment management system has its own specific interface specifications, including data format and transmission protocol requirements. To ensure that the monitoring instructions can be correctly received and processed by the equipment management system, the above information needs to be encapsulated according to the interface specifications. For example, the maintenance priority ranking, component identification, and maintenance guidance information are organized into a specific data structure, such as JSON or XML format, and encapsulated according to the transmission protocol specified by the equipment management system. The encapsulated monitoring instructions contain all the key information required for maintenance work and meet the receiving requirements of the equipment management system. Finally, the generated monitoring instructions are sent to the equipment management system. Upon receiving the monitoring instructions, the equipment management system can trigger the corresponding maintenance response mechanism, arrange maintenance personnel to perform maintenance on the equipment, and thus ensure that the mechanical parking equipment can be restored to normal operation as soon as possible.

[0163] Figure 2 The illustration shows exemplary hardware and software components of an IoT-based smart parking equipment monitoring system 100 that can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used in the IoT-based smart parking equipment monitoring system 100 and to perform the functions described in this application.

[0164] The IoT-based intelligent monitoring system for parking equipment 100 can be a general-purpose server or a special-purpose server; both can be used to implement the IoT-based intelligent monitoring method for parking equipment of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0165] For example, the IoT-based intelligent parking equipment monitoring system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the IoT-based intelligent parking equipment monitoring system 100 may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The IoT-based intelligent parking equipment monitoring system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0166] For ease of explanation, only one processor is described in the IoT-based intelligent parking equipment monitoring system 100. However, it should be noted that the IoT-based intelligent parking equipment monitoring system 100 of this application may also include multiple processors. Therefore, the steps performed by one processor as described in this application may also be performed jointly or individually by multiple processors. For example, if the processor of the IoT-based intelligent parking equipment monitoring system 100 performs steps A and B, it should be understood that steps A and B may also be performed jointly by two different processors or individually by one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0167] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned intelligent supervision method for parking equipment based on the Internet of Things is implemented.

[0168] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A smart monitoring method for parking equipment based on the Internet of Things, characterized in that, The method includes: Acquire a multi-source monitoring data set of mechanical parking equipment collected in real time by an Internet of Things sensor network. The multi-source monitoring data set includes displacement data of the lifting mechanism, power data of the lateral movement mechanism, and status data of the locking mechanism over a continuous period of time. Feature extraction is performed on the multi-source monitoring data set to generate structural stability features and motion smoothness features of the equipment's mechanical characteristics. The structural stability features reflect the temporal consistency of the geometric shape of key components, and the motion smoothness features reflect the continuity of the timing of the execution actions. The structural stability features and the motion smoothness features are input into a pre-trained anomaly recognition model for anomaly evaluation, generating a device status evaluation result that includes anomaly presence identifiers and information about anomaly-related components. Based on the abnormal associated component information in the equipment status assessment results, the characteristic sequences of the corresponding abnormal associated components in historical time periods are extracted and time-series comparison analysis is performed to determine the scope of abnormal impact in the mechanical parking equipment. Based on the comparison results of the anomaly's impact range and historical feature sequences, a regulatory instruction containing maintenance priority and component identification is generated, and the regulatory instruction is sent to the equipment management system to trigger a maintenance response; The step of extracting features from the multi-source monitoring data set to generate structural stability features and smoothness features of the equipment's mechanical characteristics includes: The multi-source monitoring data set is timestamped to obtain synchronous monitoring data; The displacement data of the lifting mechanism in the synchronous monitoring data is subjected to time-series fluctuation analysis, and the variance parameter of the displacement value in a continuous time period is calculated as the first component of the structural stability characteristic. The power data of the transverse mechanism in the synchronous monitoring data is decomposed by frequency components, and the frequency concentration parameter of the power value in the action cycle is extracted as the first component of the action smoothness feature. The frequency concentration parameter is the ratio of the energy proportion of the main frequency component to the energy proportion of the secondary frequency component in the frequency distribution diagram of the power value. The state duration statistical processing is performed on the locking mechanism state data in the synchronous monitoring data, and the standard deviation parameter of the locking state holding time is calculated as the second component of the structural stability feature. The dynamic data of the lateral movement mechanism and the displacement data of the lifting mechanism in the synchronous monitoring data are subjected to correlation and coupling analysis, and the correlation coefficient parameter between the dynamic value and the displacement value is extracted as the second component of the motion smoothness feature. The first and second components of the structural stability feature are weighted and fused to generate the final structural stability feature. The first and second components of the motion smoothness feature are weighted and fused to generate the final motion smoothness feature.

2. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 1, characterized in that, The step of performing time-series fluctuation analysis on the displacement data of the lifting mechanism in the synchronous monitoring data, and calculating the variance parameter of the displacement value within a continuous time period as the first component of the structural stability characteristic, includes: The displacement value sequence of the lifting mechanism in a continuous action cycle is extracted from the synchronous monitoring data. The displacement value sequence includes the initial displacement value at the start time and the final displacement value at the end time of each action cycle. The displacement value sequence is divided into sliding windows, and each sliding window contains a preset number of continuous action cycles; Calculate the average value of the initial displacement and the average value of the final displacement within each sliding window, and use them as the reference initial displacement and reference final displacement of that sliding window, respectively. Calculate the average of the squared differences between the initial displacement value and the reference initial displacement for each action cycle within the sliding window, and use this as the initial displacement fluctuation variance. Calculate the average of the squared differences between the termination displacement value and the reference termination displacement for each action cycle within the sliding window, and use this as the variance of the termination displacement fluctuation. The sum of the initial displacement fluctuation variance and the final displacement fluctuation variance is used as the displacement value variance parameter of the sliding window. The displacement value variance parameter is used to characterize the positional stability of the lifting mechanism within a continuous action cycle.

3. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 1, characterized in that, The step of performing frequency component decomposition processing on the lateral movement mechanism power data in the synchronous monitoring data, and extracting the frequency concentration parameter of the power value within the action cycle as the first component of the action smoothness feature, includes: The power value sequence of the traverse mechanism within a single complete action cycle is extracted from the synchronous monitoring data. The power value sequence includes the power values ​​of the action start-up phase, action execution phase, and action stop phase. The dynamic numerical sequence is processed by Fast Fourier Transform to generate a frequency distribution map of the dynamic values, which includes the energy intensity corresponding to different frequency components. Identify the dominant frequency component with the highest energy intensity and its corresponding energy percentage in the frequency distribution diagram; The ratio of the energy proportion of the main frequency component to the energy proportion of the secondary frequency component is calculated and used as the frequency concentration parameter of the dynamic value. The frequency concentration parameters of multiple consecutive action cycles are exponentially smoothed to generate a smooth concentration parameter that reflects the stability of the lateral movement frequency. The smooth concentration parameter is used to characterize the smoothness of the lateral movement mechanism's action execution.

4. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 1, characterized in that, The step of inputting the structural stability features and the motion smoothness features into a pre-trained anomaly recognition model for anomaly evaluation, and generating a device status evaluation result containing anomaly presence identifiers and information on anomaly-related components, includes: The structural stability features and the motion smoothness features are input into the input layer of the anomaly detection model, and the structural stability features and the motion smoothness features are subjected to standardized feature vector mapping processing to generate standardized feature vectors. The standardized feature vectors are subjected to pairwise correlation analysis through the feature association layer of the anomaly recognition model to generate an association feature matrix that reflects the interaction between structural stability and motion smoothness. The temporal memory layer of the anomaly detection model is used to perform historical data matching processing on the associated feature matrix to extract the similarity measure between the current associated features and the historical normal associated features; The similarity metric is threshold-judged by the decision output layer of the anomaly identification model. If the similarity metric is lower than the preset threshold, an anomaly presence flag is output. The responsibility attribution module of the anomaly identification model is used to locate the feature component that contributes the most to the decrease in the similarity metric and determine the corresponding anomaly-related component information. If the similarity metric is higher than or equal to the preset threshold, a normal presence flag is output.

5. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 4, characterized in that, The step of using the temporal memory layer of the anomaly detection model to perform historical data matching processing on the associated feature matrix and extracting similarity metrics between the current associated features and historical normal associated features includes: Extract historical normal association feature sequences that are consistent with the current time window length from the historical database of the anomaly identification model; The current associated feature matrix and the historical normal associated feature sequence are input into a dynamic time warping algorithm for sequence alignment processing to generate an aligned feature matching path. Calculate the sum of the Euclidean distances between the corresponding feature values ​​on the matching path, and use it as the difference parameter between the current associated feature and the historical normal associated features; The difference parameter is input into the similarity conversion function to generate a similarity metric. The larger the similarity metric, the more similar the current associated feature is to the historical normal associated feature.

6. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 1, characterized in that, The step of extracting the characteristic sequences of the corresponding abnormally associated components in historical time periods based on the abnormal component information in the equipment status assessment results and performing time-series comparison analysis to determine the scope of abnormal impact in the mechanical parking equipment includes: Based on the information of the abnormal associated component, the feature sequence of the abnormal associated component within a preset time period is extracted from the historical storage records of the multi-source monitoring data set. The feature sequence includes structural stability feature components and motion smoothness feature components. The historical feature sequence is divided into sliding windows, with each sliding window containing the same number of consecutive time points; Calculate the average value of the feature values ​​within each sliding window as the baseline feature value for that sliding window; Calculate the absolute difference between the feature value of the abnormal associated component and the baseline feature value of each sliding window in the current time period, and identify the sliding window with the largest absolute difference as the abnormal starting associated window; Extract the time range corresponding to the abnormal initiation correlation window, and combine it with the correlation analysis of the equipment's mechanical structure to determine the changes in the characteristics of other components affected by the abnormal component within this time range; Based on the changes in the characteristics of the other components, a description of the scope of the abnormal impact, including the abnormal components and the affected components, is generated.

7. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 6, characterized in that, The process of generating an abnormal impact range description, including the abnormal components and affected components, based on the changes in the characteristics of the other components, includes: The characteristic change values ​​of the other components are subjected to threshold judgment processing, and the components whose characteristic change values ​​exceed the preset impact threshold are identified as directly affected components. A secondary threshold judgment process is performed on the feature change values ​​of the directly affected components to identify the components whose feature changes are caused by the state changes of the directly affected components as indirectly affected components. The number and distribution location of the abnormal components, directly affected components, and indirectly affected components are recorded. Based on the quantity and distribution location, an abnormal impact range description is generated, which includes a component identification list and a spatial distribution diagram. The component identification list contains unique identifiers for abnormal components, directly affected components, and indirectly affected components. The spatial distribution diagram reflects the relative positional relationship of each component in the parking equipment.

8. The intelligent monitoring method for parking equipment based on the Internet of Things according to claim 1, characterized in that, Based on the comparison results of the anomaly's impact range and historical feature sequences, a regulatory instruction containing maintenance priority and component identification is generated, including: The fault impact degree assessment is performed on the abnormal components, directly affected components, and indirectly affected components within the scope of the abnormality. The fault impact degree assessment is based on the functional importance of each component during equipment operation and the magnitude of characteristic changes. Based on the assessment results of the degree of impact of the fault, the maintenance priorities of each component are ranked, with components that are functionally important and have large characteristic change values ​​having higher priority. Extract the component identifiers and corresponding priority levels from the maintenance priority sorting; Based on the comparison results of the historical feature sequences, the abnormal start time and feature change trend are combined to generate maintenance guidance information that includes component maintenance sequence and maintenance content suggestions; The maintenance priority sorting, component identification, and maintenance guidance information are formatted and encapsulated to generate regulatory instructions that conform to the interface specifications of the equipment management system.

9. A smart monitoring system for parking equipment based on the Internet of Things, characterized in that, The IoT-based intelligent monitoring system for parking equipment includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the memory to implement the IoT-based intelligent monitoring method for parking equipment as described in any one of claims 1-8.

Images