Automatic monitoring electric protective cover remote starting method and system of total station

Abstract

The present application relates to total station automatic monitoring technical field, disclose a kind of total station automatic monitoring electric protective cover remote starting method and system, the method includes obtaining remote wake-up instruction acquisition current to obtain first current sequence, remove outlier to obtain second current sequence;Extract time-domain current amplitude difference and frequency energy distribution characteristics fusion as multidimensional feature vector;Input support vector machine model determines preliminary running stage, for uniform speed rotation stage time reading internal state register verifies self-check, self-check is completed and generates protective opening instruction;According to protective opening instruction control motor executes mechanical action, parses position encoder signal and obtains feedback value, feedback value is lower than preset feedback threshold and mechanical action completion time falls into wake-up rhythm time window, then determine timing match success, terminate blind instruction issuing mechanism, establish measurement preparation state.The present application can solve the problem that the existing technology exists that protective cover starting time and total station wake-up rhythm are difficult to match.

Description

Technical Field

[0001] This invention relates to the field of total station automated monitoring technology, and in particular to a method and system for remotely starting the electric protective cover of a total station automated monitoring system. Background Technology

[0002] Currently, in the field of total station automated monitoring technology, total stations are crucial for ensuring the safety of engineering structures by recording point displacements unattended for extended periods in deformation observation of bridges, tunnels, slopes, and large structures. These monitoring stations are typically built outdoors in remote or high-altitude locations, powered by solar panels and batteries, and equipped with environmental sensing devices such as temperature sensors to monitor operating conditions. The equipment operates in a low-power sleep mode during observation breaks, and is remotely woken up and its electrically operated protective cover activated before the start of a new task.

[0003] In one existing technology, the remote start-up of the protective cover uses a timed, blind-sending method. After receiving the observation task, the remote end sends power-on and cover-opening commands at preset intervals. The underlying controller inside the equipment also sends default detection commands at a fixed frequency. The two-layer mechanism of timed re-sending from the remote end and blind-sending from the underlying controller operates in tandem. It does not sense whether the total station is in sleep, power-on, or self-test phase, nor does it verify whether the mechanical movements of the protective cover match the equipment status. The wake-up chain of a total station in deep sleep is long, and the time consumption of each link varies, which cannot be adapted to fixed-time commands. If the protective cover is opened before the equipment has completed its self-test, the first few sets of observation data will be invalid or disordered; if it is opened too late, the observation window will be missed, and even interference may occur between the motor movement and the telescope rotation. Since the total station does not report the start-up progress to the outside, the remote end can only resend commands multiple times to align the timing, which increases the power supply burden and causes control chaos.

[0004] Existing technologies suffer from low accuracy in matching the remote start-up timing of the protective cover with the total station's wake-up rhythm. Summary of the Invention

[0005] This invention provides a method and system for remotely activating the electric protective cover of a total station for automated monitoring, in order to solve the problem in the prior art that the activation timing of the protective cover is difficult to match with the wake-up rhythm of the total station.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides a remote start method for the electric protective cover of a total station's automated monitoring system, comprising:

[0007] Obtain the remote wake-up command and wake up the total station, collect the power supply current during the wake-up process, and obtain the first current sequence;

[0008] The first current sequence is subjected to outlier removal processing to remove electromagnetic interference spike data, resulting in the second current sequence;

[0009] Based on the second current sequence, the time-domain current amplitude difference and frequency-domain energy distribution features are extracted and fused into a multi-dimensional feature vector;

[0010] The multidimensional feature vector is input into a pre-trained support vector machine model for classification processing. The preset feature-stage mapping relationship is compared to determine the initial operation stage of the total station. When the initial operation stage is the uniform rotation stage, the internal status register is read to verify the self-test status. If it is the self-test completion stage, a protection activation command is generated. When the initial operation stage is the initial excitation stage or the self-test status is the incomplete self-test stage, the power supply current is collected and the multidimensional feature vector is updated.

[0011] According to the protection activation command, the motor is controlled to perform mechanical actions. The position encoder signal during the operation of the motor is collected and analyzed to obtain the equipment status feedback value. If the equipment status feedback value is lower than the preset feedback threshold and the completion time of the mechanical action falls within the time window corresponding to the total station wake-up rhythm, it is determined that the timing of the mechanical action and the total station wake-up rhythm is successfully matched. The mechanism of the underlying controller periodically issuing the default detection command according to the blind transmission mechanism is terminated, and the measurement preparation state is established.

[0012] Secondly, the present invention provides a remote start system for an automated monitoring electric protective cover of a total station, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0013] Compared with the prior art, the present invention has the following beneficial effects:

[0014] (1) The present invention performs abnormal value removal processing on the power supply current during the wake-up process, removes the spike data generated by environmental electromagnetic interference, provides a reliable current data basis for subsequent operation stage identification, and reduces the impact of stage misjudgment caused by electromagnetic interference.

[0015] (2) Based on the fusion of time-domain current amplitude difference and frequency-domain energy distribution characteristics into a multi-dimensional feature vector and after classification by support vector machine, the invention reads the internal status register to verify the self-test status, transforming the start-up progress that the total station does not report to the outside into the quantitative identification of current characteristics and local register reading. Without the need to add a status reporting channel on the equipment side, the operating stage and self-test completion status can be grasped, providing a basis for judgment on when to open the protective cover.

[0016] (3) The present invention performs timing matching between the completion time of the protective cover opening action and the time window corresponding to the total station wake-up rhythm, and terminates the blind command mechanism only when the matching is successful. This reduces the impact of the protective cover opening too early causing abnormal initial observation data or opening too late missing the observation window, and improves the accuracy of matching the protective cover start-up timing with the equipment wake-up rhythm. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of a method for remotely starting an electric protective cover for automated monitoring of a total station, provided in the first embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Reference Figure 1 The first embodiment of the present invention provides a method for remotely starting the electric protective cover of a total station for automated monitoring, comprising the following steps:

[0020] S1, Obtain the remote wake-up command and wake up the total station, collect the power supply current during the wake-up process, and obtain the first current sequence;

[0021] S2, perform outlier removal processing on the first current sequence to remove electromagnetic interference spike data and obtain the second current sequence;

[0022] S3, extract the time-domain current amplitude difference and frequency-domain energy distribution features from the second current sequence, and fuse them into a multi-dimensional feature vector;

[0023] S4, input the multidimensional feature vector into the pre-trained support vector machine model for classification processing, compare the preset feature and stage mapping relationship, determine the initial operation stage of the total station, when the initial operation stage is the uniform rotation stage, read the internal status register to verify the self-test status, if it is the self-test completion stage, generate the protection activation command, when the initial operation stage is the initial excitation stage or the self-test status is the incomplete self-test stage, return to collect the power supply current and update the multidimensional feature vector;

[0024] S5. Control the motor to perform mechanical actions according to the protection activation command, collect and analyze the position encoder signal during the operation of the motor, obtain the equipment status feedback value. If the equipment status feedback value is lower than the preset feedback threshold, and the completion time of the mechanical action falls within the time window corresponding to the total station wake-up rhythm, it is determined that the timing of the mechanical action and the total station wake-up rhythm is successfully matched. The mechanism of the underlying controller periodically issuing the default detection command according to the blind transmission mechanism is terminated, and the measurement preparation state is established.

[0025] In step S1, a remote wake-up command is obtained and the total station is woken up. The power supply current during the wake-up process is collected to obtain the first current sequence, including:

[0026] According to the remote wake-up command, the power supply circuit of the total station is triggered to close;

[0027] Acquire the initial voltage signal of the power supply circuit. If the initial voltage signal is greater than a preset voltage threshold, collect the power supply current of the total station during the transition from sleep state to active state.

[0028] The supply current is continuously sampled to obtain the first current sequence.

[0029] First, the remote wake-up command is issued by the remote monitoring center before the observation task begins. This triggers an internal relay via the wireless communication module, closing the total station's power supply circuit. During observation intervals, the total station is in a low-power deep sleep state. Upon receiving the command, the main control unit first activates itself and then supplies power to the entire unit, eliminating the need for personnel to be present on-site.

[0030] Secondly, voltage verification is performed after the power supply circuit is closed. The preset voltage threshold is determined based on the undervoltage protection point of the total station's built-in battery, which is 11.5 volts in this embodiment. In the field environment, the battery capacity may decrease due to low temperature or long-term standby. Forcibly starting under undervoltage conditions can easily lead to initialization failure of internal precision optical components and motors or hardware damage. For example, when the detected initial voltage signal is 12.2 volts, it is determined that it is greater than the preset voltage threshold, and starting is allowed; if the initial voltage signal is not greater than the preset voltage threshold, it returns to reacquire. When the preset undervoltage retry limit is reached, an undervoltage alarm event is reported to the remote monitoring center through the remote communication link and the current startup process is suspended. The preset undervoltage retry limit is set to 3 times. The basis for this is that the undervoltage that occurs after the battery is at low temperature or after long-term standby is usually a temporary decrease in capacity. When the battery terminal voltage is retested after 10 seconds, it can recover due to the redistribution of internal charge. However, if the voltage is lower than the undervoltage protection point for 3 consecutive tests, it indicates that the battery is in a deep discharge state. Continuing to retest will not recover, so 3 times is set as the limit. Voltage verification and subsequent dynamic current monitoring work together to construct the current timing sequence after voltage confirmation.

[0031] Then, the power supply current of the total station during the transition from sleep to active state is collected. This transition process accompanies the power-on of each submodule, and the current characteristics are highly dynamic. The current is discretized using an analog-to-digital converter at a preset sampling frequency. The preset sampling frequency is determined based on the bandwidth of the transient process in the current signal; in this embodiment, 1000Hz is used, which can fully capture the millisecond-level current jump at power-on without generating redundant data. During sampling, a moving average filter can be used to remove high-frequency noise, ultimately constructing time-series data, namely the first current sequence. This sequence records the complete dynamic trajectory from the microampere-level standby current to the ampere-level operating current.

[0032] In step S2, outlier removal processing is performed on the first current sequence to remove electromagnetic interference spikes, resulting in a second current sequence, including:

[0033] The first current sequence is segmented by a sliding window to construct multidimensional feature space data;

[0034] The average path length of the multidimensional feature space data is calculated using the isolated forest algorithm. Anomaly scores are calculated based on the average path length. Data segments with anomaly scores greater than a preset anomaly score threshold are identified as electromagnetic interference spike data.

[0035] The electromagnetic interference spike data in the first current sequence is removed, and the sequence after removal is interpolated, filled, and smoothed to obtain the second current sequence.

[0036] First, the first current sequence is segmented using a sliding window mechanism. The window length and step size are determined based on the duration of the transient current characteristics. In this embodiment, the window length is set to 50 sampling points, and the step size is 10 sampling points. By segmenting, the one-dimensional time series is converted into a feature matrix containing multiple 50-dimensional vectors. Each vector represents the current change pattern within a very short time segment, i.e., multi-dimensional feature space data.

[0037] Secondly, the multi-dimensional feature space data is input into the constructed isolated forest model. The isolated forest model is pre-trained offline. During the sample collection phase, power supply current data is collected during multiple normal startups of the device in an environment free from electromagnetic interference. Each startup forms a complete first current sequence sample. During the sample processing phase, each first current sequence sample undergoes the same sliding window segmentation as online detection, with the window length and step size consistent with online detection. Each sample is segmented into multiple 50-dimensional vectors, which are then aggregated to generate a multi-dimensional feature space training sample set. During the model training phase, for each isolated tree, a fixed-size subsample is randomly selected from the training sample set. Random feature selection and random segmentation values ​​are recursively performed on these subsamples. Each segmentation divides one feature dimension into two subspaces according to a random segmentation value, until each sample point is isolated or the preset tree height is reached. The subsample size and preset tree height are set based on conventional experience. After training, the constructed isolated tree set and normalized parameters are stored for direct use during online detection.

[0038] During online detection, the multi-dimensional feature space data obtained from real-time segmentation is input into the trained Isolation Forest model. For each data point, the path length from the root node to the leaf node in each isolated tree is calculated and averaged to obtain the average path length. The shorter the average path length, the easier it is for the data point to be isolated, the more it deviates from the normal fluctuation trajectory, and the more likely it is to be an outlier. Isolation Forest constructs isolated trees by randomly selecting features and random segmentation values. Outliers, because they deviate from the normal fluctuation trajectory, are isolated with fewer segmentations. In this embodiment, the number of isolated trees is set to 100, and the subsample size is constant. Calculate the average path length of each data point across all isolated trees. Anomaly scores are calculated based on the average path length and a pre-defined normalization function, which incorporates a normalization term related to the subsample size. The formula for its calculation is:

[0039] ;

[0040] in For the harmonic series, its approximate calculation formula is:

[0041] ;

[0042] for Calculated , The normalization term maps different path lengths to a comparable interval of 0 to 1. The anomaly score formula is as follows:

[0043] ;

[0044] It is worth noting that the preset anomaly score threshold is determined through offline statistical methods. Current samples from multiple normal starts of the device in an environment free from electromagnetic interference are collected, and the anomaly score for each sample point is calculated to obtain the score distribution of historical normal samples. The 95th percentile of this distribution is taken as the threshold, which is 80 in this example. Data segments with anomaly scores greater than the preset anomaly score are identified as containing spikes caused by relay contact jitter or external electromagnetic interference.

[0045] Finally, the marked spike data points are located and removed from the first current sequence. For data segments identified as spikes, an overlapping window voting mechanism is used to locate the spike data points. When the same sample point falls into multiple overlapping windows, the number of times it is identified as an anomaly in each window is counted. When the number of windows identified as an anomaly exceeds half of the total number of windows in which it appears, the sample point is identified as a spike data point. Only the located spike data points are removed. For the missing positions left after removal, cubic spline interpolation is used to fill in the gaps, and the data is reconstructed using the smooth transition characteristics of the current values ​​before and after the missing positions. After filling in the gaps, Gaussian filtering is applied to smooth the entire sequence, with the Gaussian kernel window size set to 5 and the standard deviation to 2.

[0046] The output second current sequence removes transient abnormal changes while retaining the low-frequency current change trend during the activation process. The parameter settings are based on the fact that a 50-point window covers a 50ms time interval at a 1000Hz sampling frequency, which is sufficient to cover the duration of the relay jitter spike. A Gaussian kernel window of 5 and a standard deviation of 2 suppress glitches without damaging the rise edge shape of the startup current.

[0047] In step S3, the time-domain current amplitude difference and frequency-domain energy distribution features are extracted from the second current sequence and fused into a multi-dimensional feature vector, including:

[0048] Calculate the current amplitude difference between adjacent sampling points in the second current sequence to obtain the time-domain current amplitude difference sequence;

[0049] The second current sequence is subjected to Fourier transform to obtain a spectrum sequence. The spectrum sequence is divided into multiple frequency band intervals, and the energy proportion in each frequency band interval is calculated to obtain the frequency domain energy distribution characteristics.

[0050] The time-domain current amplitude difference sequence is subjected to dimensionality reduction processing to obtain dimensionality reduction features, which are then concatenated with the frequency-domain energy distribution features to obtain fluctuation features;

[0051] The fluctuation feature is filled with zero values ​​up to a preset dimension threshold to obtain the multidimensional feature vector.

[0052] First, calculate the difference in current amplitude between two adjacent sampling points in the second current sequence, i.e., the difference in current values ​​between adjacent sampling points, in amperes (A). Taking a second current sequence with a fixed length of 1000 sampling points (corresponding to a standard sampling duration of 1 second) as an example, zeros are padded at the end when the length is insufficient, and the first 1000 sampling points are truncated when the length is excessive. After first-order difference calculation, a current amplitude difference sequence with a fixed length of 999 is output. This sequence intuitively reflects the severity of the transient current rise. The difference is large and dense during the excitation stage, and small and sparse during the stable standby stage.

[0053] Secondly, the second current sequence is input into a Fast Fourier Transform algorithm to convert the time-domain current signal to the frequency domain, outputting a spectral sequence. This spectral sequence reveals the characteristic frequency distribution generated by different electronic components inside the total station during operation. The spectral sequence is divided into frequency bands, with the band boundaries determined based on the operating frequency characteristics of the components inside the total station. In this embodiment, the low-frequency range is set to 0 to 50 Hz, the mid-frequency range to 50 to 200 Hz, and the high-frequency range to 200 to 500 Hz. The low-frequency and mid-frequency ranges are semi-open ranges with the left side closed and the right side open, i.e., the low-frequency range [0, 50), the mid-frequency range [50, 200), and the high-frequency range is a closed range [200, 500]. The unit is Hz. Each frequency point is uniquely assigned to one range to avoid duplicate counting of the range boundaries. The DC component is counted separately and does not participate in the percentage calculation. The Nyquist frequency point is included in the high-frequency range. Using a single-sided spectrum, the DC component and Nyquist frequency are not multiplied, while the amplitudes of other frequency points are multiplied by 2. The sum of the squares of the spectral amplitudes in the three intervals is calculated, divided by the total energy of the three AC frequency bands after deducting the DC component, and multiplied by 100. The output includes energy distribution characteristics containing three percentage values. A high proportion of low frequencies indicates that the motor drive has not yet started, while an increase in the proportion of mid-to-high frequencies indicates that the servo motor has entered the excitation and rotation stage.

[0054] Then, due to the high dimensionality of the current amplitude difference sequence, principal component analysis (PCA) was used for dimensionality reduction. The number of principal components retained was determined based on a cumulative contribution rate of 85%, and the top 10 principal components were determined through offline verification using historical samples. A fixed 999-dimensional current amplitude difference sequence was input, and 10-dimensional reduced features were output. The 10-dimensional reduced features were linearly concatenated with the 3-dimensional energy distribution features to construct a fluctuation feature with a total dimension of 13.

[0055] Finally, to meet the consistency requirements of the input dimensions for subsequent pattern recognition models, a preset dimension threshold of 20 is set. The current fluctuation feature dimension is fixed at 13, which is lower than the preset threshold of 20. Seven zero values ​​are appended to the end of the fluctuation feature sequence for padding, ultimately outputting a multi-dimensional feature vector of fixed length 20 dimensions. Dimensions 1 to 10 are the current amplitude difference features after PCA dimensionality reduction; dimensions 11 to 13 are the frequency domain energy distribution features; dimensions 14 to 18 are zero-value filling slots; and dimensions 19 and 20 are reserved overwrite slots. During subsequent online updates, dimensions 19 and 20 are overwritten by the frequency domain amplitude features. These frequency domain amplitude features are the feature components of the current fluctuation features extracted through frequency domain analysis and overwritten to the 19th and 20th dimension slots as described in step S4. These two are corresponding representations of the same object at different process stages. The dimensionality reduction matrix of principal component analysis is calculated offline from historical current samples under standard sampling duration during the training phase and pre-stored in the device storage area. During the inference phase, the pre-stored matrix is ​​directly called to perform projection operations, avoiding the drift of dimensionality reduction results due to differences in sampling duration and ensuring that the input feature space of the support vector machine model remains consistent.

[0056] In step S4, the multidimensional feature vector is input into a pre-trained support vector machine model for classification processing. The preset feature-stage mapping relationship is compared to determine the initial operating stage of the total station. When the initial operating stage is a uniform rotation stage, the internal status register is read to verify the self-test status. If it is a self-test completed stage, a protection activation command is generated. When the initial operating stage is an initial excitation stage or the self-test status is an incomplete self-test stage, the system returns to collecting the power supply current and updating the multidimensional feature vector, including:

[0057] The multidimensional feature vectors are input into a support vector machine model for mapping to obtain mapped feature data.

[0058] Calculate the signed decision function value corresponding to the mapped feature data to obtain the decision value, perform a signed function operation on the decision value, and output the classification label;

[0059] Search for the stage identifier corresponding to the classification label in the preset feature-stage mapping relationship, and determine the initial operation stage based on the stage identifier;

[0060] The running stage code in the internal status register is compared with a preset status mapping table to obtain a self-test status identifier, which includes the self-test completion stage and the self-test incomplete stage.

[0061] When the self-test status is marked as the self-test completion stage, the current external ambient temperature data and wind speed data of the total station are extracted, and the external ambient temperature data and wind speed data are input into a preset instruction encoding rule library to generate the protection activation instruction containing environmental parameters.

[0062] When the self-test status is marked as the incomplete self-test stage, the system returns to collect the power supply current and update the multidimensional feature vector. If the self-test completion stage is still not found after the preset monitoring time limit is reached, the system reports a self-test timeout event through the remote communication link and terminates the current startup process.

[0063] First, a multi-dimensional feature vector of length 20 is obtained. The mean and standard deviation vectors saved during the training phase are then used to perform the same Z-score standardization on the current feature vector, ensuring that the feature distribution during the inference phase is consistent with that during training. This vector is then input into the pre-trained support vector machine model. The radial basis function (RBF) implicitly projects the feature vectors, originally in 20-dimensional space, to a higher-dimensional feature space, making the features from different operational phases separable in the higher-dimensional space. The RDF kernel function has the following form:

[0064] ;

[0065] in, Given the current input multidimensional feature vector, For support vectors, The kernel function bandwidth parameter is determined by cross-validation during the training phase. for and Euclidean distance, It is an exponential function. The kernel function implicitly maps the 20-dimensional features to a higher-dimensional space, and then calculates the signed decision function value corresponding to the mapped feature data:

[0066] ;

[0067] in, Given the current input multidimensional feature vector, For the i-th support vector, for The corresponding Lagrange multipliers, for Category label for the stage (+1 for uniform rotation stage, -1 for initial excitation stage). Here, b is the kernel function, and b is the bias term. The sign determines the classification label, and the absolute value reflects the classification confidence level. The decision value is then obtained.

[0068] A preset decision confidence threshold of 2 is set, determined based on the statistical lower bound of the support vector machine (SVM) classification margin. In the SVM model, the width of the classification margin is 2 divided by... ,in The hyperplane normal vectors are processed by L2 normalization after training to make them... The norm is 1, corresponding to a theoretical classification margin width of 2. Decision function value The absolute value of reflects the functional distance from the sample point to the hyperplane. When the absolute value is greater than or equal to 1, the sample falls outside the margin boundary and the classification confidence is high. When the absolute value is less than 1, the sample falls within the interval, indicating low classification confidence. This embodiment statistically analyzes the number of correctly classified samples from historical launch samples. The absolute value distribution shows that the 5th percentile is approximately 2, meaning that about 95% of the correctly classified samples have this property. Since the absolute value is greater than 2, 2 is used as the preset decision confidence threshold to filter out fuzzy samples that fall near the classification boundary.

[0069] If the calculated decision value is greater than the preset decision confidence threshold of 2, satisfying the classification trigger condition, the decision value is input into the sign function for calculation. The sign function discretizes the input value based on its positive or negative attribute; it outputs the integer 1 when the input is positive and -1 when the input is negative. For example, a decision value of 4 outputs a classification label of 1 after processing by the sign function; conversely, a decision value of -4 outputs a classification label of -1. If the absolute value of the decision value is not greater than the preset decision confidence threshold, it indicates that the sample falls in the transition region near the hyperplane. In this case, no definitive label is output, and the process returns to continue collecting power supply current and updating the feature vector.

[0070] It is worth noting that the preset feature-stage mapping relationship library stores the correspondence rules between discrete labels and the physical operating status of the total station. Label 1 is set to correspond to the initial excitation stage of the total station, and label -1 corresponds to the uniform rotation stage of the total station. Based on the currently output classification label, a table lookup and matching are performed to extract the corresponding stage identifier and determine the initial operating stage of the total station.

[0071] The support vector machine (SVM) model is a binary classification model, pre-trained. Power supply current data was collected from at least 10 total stations of the same model during multiple normal wake-up processes within a temperature range of -10℃ to 40℃. Multidimensional feature vector samples were generated through the feature extraction process described in steps S1 to S3. The total training sample size was 120 sets, with 60 sets each for the initial excitation stage and the uniform rotation stage. The collected multidimensional feature vectors were used as input, and manually labeled initial operation stages were used as labels. Supervised offline training was employed. The label boundaries of the training samples were defined by whether the motor speed reached the preset stable speed. Samples from motor start-up until the speed first reaches the preset stable speed are uniformly labeled as the initial excitation stage (label 1). Samples after the speed reaches and maintains the preset stable speed are uniformly labeled as the uniform rotation stage (label -1). The preset stable speed is determined based on the rated operating speed of the total station servo motor. In this embodiment, it is 60 r / min. That is, when the motor speed first reaches and stabilizes at 60 r / min, the samples after the corresponding time point are labeled as the uniform rotation stage label -1. The final determination of whether the self-test is completed is verified by reading the status register later, and does not depend on the output of this model.

[0072] Because the current amplitude difference characteristics (in A), frequency domain energy distribution characteristics (dimensionless percentage), and spectral amplitude characteristics (in A) after PCA dimensionality reduction in a 20-dimensional vector have different dimensions and orders of magnitude, directly in Hybrid computation can lead to distance metrics being dominated by a large number of features. Therefore, during the training phase, the mean and standard deviation of 120 training samples are calculated for each dimension. Z-score standardization is then performed on the 20-dimensional vector to convert each feature into a dimensionless value with zero mean and unit variance. For zero-valued filler slots in dimensions 14 to 18, which are always zero and have a standard deviation of zero in the training samples, to avoid division by zero in the standardization operation, these zero-valued filler slots are not included in the standardization and distance calculation. They are directly kept as zero-valued inputs, which is equivalent to them not contributing any component to the kernel function distance metric.

[0073] This support vector machine model is a binary classification model, using a radial basis function kernel as the kernel function, and the kernel parameters are... and regularization parameters All values ​​were determined using five-fold cross-validation based on historical starting samples. The grid of combinations where γ was in the [0.1, 10] interval and C was in the [0.1, 10] interval was traversed, and the parameter combination that achieved the highest classification accuracy was selected as the final value. After cross-validation, γ=0.5 and C=1 were selected. The 120 standardized training samples and their corresponding labels were input into the SVM model, and sequential minimum optimization was used to solve for the optimal classification hyperplane to complete model training. After training, the model parameters, as well as the mean and standard deviation vectors used for standardization, were saved for use during the inference phase.

[0074] It should be noted that the support vector machine model in this scheme is responsible for the initial operational phase identification at the current characteristic level. Determining whether the current operational phase belongs to the self-test completion stage is not directly based on the classification label. Instead, it uses the classification label indicating that the equipment has entered the uniform rotation stage as a trigger condition, and then reads the internal status register of the total station's main control chip to verify the self-test status. When the classification label is -1 (uniform rotation stage), a query operation between the internal status register and the status mapping table is triggered, and the self-test status identifier obtained from the query is used as the final criterion. When the classification label is 1 (initial excitation stage), it returns to continue collecting the power supply current and updating the feature vector, avoiding premature activation of the protective shield due to misjudging the self-test progress solely based on current characteristics.

[0075] When the initial operation phase is the initial excitation phase, it indicates that the total station is still in the early power-on process. It returns to collecting the power supply current and updating the multidimensional feature vector, waiting for the next round of classification and judgment. Only when the initial operation phase is the uniform rotation phase will the reading of the internal status register of the main control chip of the total station be triggered to verify the self-test status.

[0076] Specifically, the main control unit accesses the register address space of the total station's main control chip via the internal bus, reads the operating stage code according to the register offset defined by the status coding specification, compares this code with a preset status mapping table, and extracts the corresponding self-test status identifier. It should be noted that in this embodiment, the main control unit and the total station's main control chip are deployed within the same device and are directly connected via the internal bus. Reading the internal status register is an internal data exchange within the device and does not rely on the total station reporting the startup progress to the remote monitoring center.

[0077] The status mapping table is pre-built based on the status coding specifications of the total station's main control chip. The mapping rule is that codes 1 to 15 correspond to each power-on initialization sub-stage, code 16 corresponds to the self-test completion stage, and codes 17 and above correspond to the measurement operation stage. It should be noted that the measurement operation stage is arranged after the self-test completion stage in the status coding sequence, indicating that the equipment self-test has passed and has entered the measurement ready state. Therefore, in determining the self-test status identifier, measurement operation stages with codes 17 and above are included in the self-test completion stage processing, i.e., triggering the protection activation command construction process. For example, reading the operation stage code 16 and looking up the table, the self-test status identifier is found to be the self-test completion stage.

[0078] Then, if the self-test status indicator obtained from the table lookup is "self-test complete," it indicates that the photoelectric ranging module and angle dial inside the total station have passed the benchmark verification. At this time, the process of constructing the protection activation command is triggered. The current external ambient temperature and wind speed data of the total station are extracted. The ambient temperature data is collected by the temperature sensor installed at the monitoring station, in °C, and the wind speed data is collected by the anemometer at the station, in m / s. These environmental parameters are used to determine the deployment speed level of the protective cover during command encoding. The command encoding rule base pre-stores the correspondence between environmental parameters and deployment speed levels. For example, when the temperature is higher than 10°C and the wind speed is lower than 5 m / s, the normal speed level is selected; when the temperature is not higher than 10°C or the wind speed is not lower than 5 m / s, the low speed level is selected. The normal speed level corresponds to a deployment speed of 5 mm / s, and the low speed level corresponds to a deployment speed of 2 mm / s, avoiding interference between the protective cover deploying too quickly and the wake-up action.

[0079] For example, temperature data of 25℃ and wind speed data of 3m / s are input into a preset command encoding rule library, matched with a normal speed setting, to generate a protection activation command containing environmental parameters. This command drives the protective cover motor, controlling the protective cover to perform an unfolding action at a speed of 5mm / s. Specifically, before the monitoring operation period, the protective cover is in a closed state, blocking the measurement optical path. After receiving the protection activation command, the protective cover moves horizontally along the guide rail, unfolds, and moves away from the total station's measurement optical path, enabling the total station to perform measurement operations. After the measurement operation is completed, the protective cover is closed again to provide physical protection for the precision optical components.

[0080] If the self-test status flag obtained from the table lookup indicates that the self-test phase has not been completed, the current monitoring mechanism is activated, and the operation of collecting the power supply current is returned to update the multi-dimensional feature vector.

[0081] In one implementation, updating the multidimensional feature vector includes:

[0082] Obtain the real-time power supply current data of the total station and extract the current data sequence within a preset current extraction window;

[0083] The current data sequence is subjected to frequency domain analysis, and the amplitude at a preset frequency in the frequency domain signal is extracted as the current fluctuation feature.

[0084] The current fluctuation feature is replaced with the feature component at a preset position in the multidimensional feature vector to complete the update of the multidimensional feature vector.

[0085] Specifically, real-time power supply current data is acquired through the high-frequency sampling channel of the total station's power management module at a preset sampling frequency of 1000Hz. A current data sequence with a preset current capture window of 50ms is extracted and input into the Fast Fourier Transform (FFT) algorithm for frequency domain analysis. A 50-point FFT with a Hanning window is used, and the amplitude is normalized according to 2|X[k]| / N (except for DC and Nyquist frequencies). The unit is A. When the preset frequency does not fall on the frequency grid, the amplitude of the nearest frequency point is taken. The amplitude at the preset frequency in the frequency domain signal is extracted as the current fluctuation feature. The preset frequency is the excitation characteristic frequency of the servo motor and its first harmonic frequency. The excitation characteristic frequency is determined by the peak position of the spectrum of the no-load starting current during the equipment calibration stage and is pre-stored in the configuration storage area. During the inference stage, the pre-stored frequency is directly called to perform amplitude extraction.

[0086] For example, the extracted fundamental frequency amplitude is 12A and the harmonic frequency amplitude is 4A. These two amplitude values ​​are replaced in the preset frequency domain amplitude slots (i.e., the 19th and 20th dimensions) of the original 20-dimensional feature vector, completing the update of the multidimensional feature vector. The first to 18th dimensions remain unchanged. The updated feature vector reflects the micro-energy consumption pattern of the total station during the self-test process in real time. This fixed slot replacement mechanism matches the training method of the support vector machine model. During the training phase, training samples are also constructed using this fixed slot replacement method, ensuring that the feature distribution of the training samples remains consistent with that of the online inference phase, avoiding the introduction of distribution shifts due to new features that could lead to classification inaccuracies.

[0087] Furthermore, if the operation of collecting power supply current continues to be performed until the preset monitoring time limit is reached and the self-test completion stage is still not found, the local loop will stop and the self-test timeout event will be reported through the remote communication link. The remote monitoring center will then transfer the process to manual intervention. The preset monitoring time limit is 120 seconds. This is based on the fact that the statistical time for the total station to wake up from sleep mode and complete the self-test is about 60 seconds in normal temperature environment and about 90 seconds in low temperature environment. The margin of 1.3 times the statistical time in low temperature environment is taken and rounded up to 120 seconds as the monitoring time limit. If the self-test completion stage is still not found after this time, it indicates that the equipment self-test process has abnormally stopped, and continuing to wait is meaningless.

[0088] In step S5, the motor is controlled to perform mechanical actions according to the protection activation command. Position encoder signals during motor operation are collected and analyzed to obtain equipment status feedback values. If the equipment status feedback value is lower than a preset feedback threshold, and the completion time of the mechanical action falls within the time window corresponding to the total station wake-up rhythm, then the timing match between the mechanical action and the total station wake-up rhythm is determined to be successful. The mechanism of the underlying controller periodically issuing default detection commands according to the blind transmission mechanism is terminated, and a measurement preparation state is established, including:

[0089] Edge detection is performed on the position encoder signal to obtain pulse sequence features;

[0090] The angular displacement increment of the motor is determined based on the pulse sequence characteristics, and the difference between the angular displacement increment and the theoretical angular displacement increment is compared to obtain the motion trajectory deviation.

[0091] If the deviation of the motion trajectory is greater than the preset deviation threshold, the action execution state corresponding to the motion trajectory deviation is analyzed, and the action execution state is mapped to a numerical resistance index according to the pre-stored state and value comparison relationship to obtain the device status feedback value. If the deviation of the motion trajectory is not greater than the preset deviation threshold, the resistance index corresponding to the normal execution state is used as the device status feedback value.

[0092] When the device status feedback value is lower than the preset feedback threshold, a timing matching identifier is generated, and the generation time of the protection activation command is recorded.

[0093] Based on the pre-stored wake-up rhythm timing table and the timing matching identifier, the transmission time window of the channel blocking signal is determined to be the time window corresponding to the total station wake-up rhythm. Combined with the generation time, the completion time of the mechanical action is verified to fall within the time window.

[0094] Within the time window, the channel blocking signal is sent, and the mechanism for the underlying controller to periodically issue default probe commands according to the blind transmission mechanism is terminated based on the channel blocking signal, and the blocking status register is activated.

[0095] Configure the data transmission link according to the communication handshake protocol in the blocking status register, and establish the measurement preparation state;

[0096] If the completion time of the mechanical action does not fall within the time window, the timing match is determined to be unsuccessful. The motor is then reset and the mechanical action is re-executed and the completion time is re-verified. If the preset retry limit is reached and the failure still occurs, a fault event of the protection mechanism is reported through the remote communication link and the mechanical action is suspended.

[0097] First, upon receiving the protection activation command, the main control unit converts the target speed parameter in the command into a pulse width modulation (PWM) signal. The correspondence between the target speed and the duty cycle is pre-determined based on motor calibration tests and stored as a lookup table. For example, the target speed corresponding to the normal speed setting is 5 mm / s, and the corresponding PWM duty cycle found in the lookup table is 60%. The main control unit generates a PWM drive signal with a 60% duty cycle and applies it to the stator windings of the brushless DC motor, driving the motor rotor to overcome mechanical friction and begin rotating. After being reduced in speed by a planetary gear reducer with a reduction ratio of 1:50, the rotational motion is converted into linear motion by a lead screw transmission mechanism with a lead of 2 mm. The reducer output shaft drives the lead screw to rotate synchronously via a 1:1 synchronous belt transmission. Each rotation of the lead screw causes the protective cover to move linearly by 2 mm, that is, each rotation of the reducer output shaft corresponds to a linear displacement of 2 mm for the protective cover. The position encoder is installed at the end of the reducer's output shaft and rotates synchronously with it. The encoder outputs 1024 pulses per revolution; therefore, one millimeter of linear displacement corresponds to 1024 divided by 2, which equals 512 pulses. The mechanical transmission gear set driving the protective cover performs the unfolding action at a speed of 5mm / s. Using this conversion factor, 5mm / s corresponds to 2560 pulses per second, and a 100ms window corresponds to a theoretical pulse frequency of 2560Hz.

[0098] Secondly, during motor operation, the photoelectric position encoder installed at the end of the reducer output shaft rotates synchronously. The grating disk inside the encoder cuts the light source, outputting orthogonal pulse electrical signals. This signal is acquired through a high-speed counter channel, and pulse data within a continuous 100ms window is extracted. An edge detection algorithm is used to identify the rising and falling edges of the pulses, and the pulse sequence characteristics are statistically obtained. The distinction between forward pulses and reverse interference pulses is determined based on the phase lead-lag relationship of the two orthogonal pulse electrical signals. During forward rotation, phase A leads phase B by 90 degrees, and during reverse rotation, phase B leads phase A by 90 degrees. Based on this, forward and reverse counts are accumulated separately.

[0099] Then, based on the pulse sequence characteristics and the inherent resolution of 1024 pulses per revolution generated by the position encoder, the angular displacement increment of the reducer output shaft within the current time window is calculated. The angular displacement increment is then compared using pulse equivalents as the unified unit of measurement. The conversion relationship between pulse equivalents and angles is 360 degrees for every 1024 pulses, meaning each pulse equivalent corresponds to 0.3515625 degrees. For example, within the current 100ms window, there are 245 forward pulses and 1 reverse interference pulse. The theoretical angular displacement increment is calculated using a conversion factor of 512 pulses per millimeter and a target speed of 5 mm / s. 100ms corresponds to a linear displacement of 0.5 mm, equivalent to 256 pulse equivalents. The actual calculated angular displacement increment corresponds to 244 pulse equivalents (i.e., 245 forward minus 1 reverse). Subtracting the two yields a trajectory deviation of 12 pulse equivalents (corresponding to 4.21875 degrees).

[0100] A preset deviation threshold of 10 pulse equivalents (corresponding to 3.515625 degrees) is set, determined based on the normal fit tolerance conversion of the protective cover transmission gear set. Since the current deviation of 12 is greater than the threshold of 10, the action execution state analysis process is triggered. The corresponding action execution state is determined by looking up a table according to the preset deviation range in which the motion trajectory deviation is located. For example, if the motion trajectory deviation of 12 falls into the range of greater than 10 and not greater than 25 pulse equivalents, the corresponding action execution state is determined to be slight jamming.

[0101] The equipment status feedback value uses a numerical resistance index to participate in subsequent timing matching and determination. The resistance index is defined as a quantitative indicator of the magnitude of the operating resistance of the protective mechanism; a larger value indicates greater operating resistance and more abnormal action. The action execution status is mapped to the resistance index based on a pre-stored relationship between status and value. The normal execution status corresponds to a resistance index range of less than 60, the slight jamming status corresponds to a resistance index range greater than or equal to 60 and less than 75, and the severe jamming status corresponds to a resistance index range greater than or equal to 75. The resistance index corresponding to the abnormal load status is assigned to the corresponding level according to the range of its actual deviation value. The slight jamming level corresponds to a motion trajectory deviation greater than 10 and not greater than 25 pulse equivalents, and the severe jamming level corresponds to a motion trajectory deviation greater than 25 pulse equivalents. For example, an abnormal load with a motion trajectory deviation of 18 pulse equivalents falls into the slight jamming deviation range and is mapped to the resistance index range corresponding to slight jamming. For example, the slight jamming status is mapped to a resistance index of 72. If the motion trajectory deviation is not greater than a preset deviation threshold, the resistance index corresponding to the normal execution status is used as the equipment status feedback value.

[0102] It should be noted that the relationship between state and value specifies the resistance index range corresponding to each action execution state. The specific value within the range is calculated by the motion trajectory deviation and motor phase current of the current time window according to a preset linear weighting rule. The linear weighting rule is that the resistance index equals the sum of the base value, the product of the deviation coefficient and the motion trajectory deviation, and the product of the current coefficient and the phase current increment. The phase current increment refers to the difference between the motor phase current in the current time window and the preset current baseline. The motion trajectory deviation is measured in pulse equivalents, and the phase current increment is measured in amperes. The base values ​​for normal execution state, slight jamming state, and severe jamming state are preset to 30, 60, and 75, respectively. The determination of these three base values ​​is based on the base value of 30 for the normal execution state as a reference benchmark. The values ​​are determined based on the average calibrated resistance measurement of the protective cover transmission mechanism within the normal fit tolerance range; the slight jamming state corresponds to a motion trajectory deviation greater than 10 pulse equivalents, and its lower limit of resistance index must be higher than the upper limit of 60 for the normal execution state, so the base value is set to 60. This value is determined based on the lower limit of the average resistance measurement when the transmission gear set experiences slight jamming; the severe jamming state corresponds to a motion trajectory deviation greater than 25 pulse equivalents, and its lower limit of resistance index must be higher than the upper limit of 75 for the slight jamming state, so the base value is set to 75. This value is determined based on the lower limit of the average resistance measurement when the transmission gear set experiences severe jamming; the three base values ​​increase sequentially and adjacent intervals do not overlap, ensuring that the resistance index intervals for different action execution states are independent of each other and avoiding confusion at the interval boundaries.

[0103] When the load abnormality is classified into the slight jamming mode, the base value of 60 is used; when classified into the severe jamming mode, the base value of 75 is used. The deviation coefficient and the current coefficient are preset to 1 and 30, respectively. The deviation coefficient of 1 is determined based on the fact that the motion trajectory deviation is measured in pulse equivalents. A single pulse equivalent corresponds to an angular deviation of 0.3515625 degrees. The resistance increment on the linear displacement of the protective cover, converted by the transmission mechanism, is approximately 1 resistance exponential unit. Therefore, the deviation coefficient is set to 1. The current coefficient of 30 is determined based on the fact that the phase current increment is measured in amperes. Historical calibration tests show that for every 1A increase in phase current, the equivalent resistance increment of the transmission mechanism is approximately 30 resistance exponential units. This proportional relationship is obtained by linear fitting of the calibration data from the motor stall test and the load step test. Therefore, the current coefficient is set to 30. The calculation results are truncated to the resistance index range corresponding to the current action execution state. The truncation rule is that the resistance index is rounded up to the nearest integer. If the calculated result for the normal execution state reaches 60 after rounding up, it is truncated to 59. If the calculated result for the slightly stuck state reaches 75 after rounding up, it is truncated to 74. If it is below 60, it is truncated to 60. If the calculated result for the severely stuck state is below 75 after rounding up, it is truncated to 75. This ensures that the truncated value still falls within the corresponding range. The resistance index calculated for the same action execution state in different time windows may be different, as long as it falls within the corresponding range, it constitutes a valid mapping for that state. For example, in the normal execution state, the resistance index is calculated as 30 + 1 × 8 + 30 × 0.4 = 50 in the first window based on a motion trajectory deviation of 8 pulses and a phase current increment of 0.4A. In the slightly stuck state, the resistance index is calculated as 60 + 1 × 12 + 30 × 0 = 72 in the current window based on a motion trajectory deviation of 12 pulses and a phase current increment of 0A. In the normal window after re-collection, the resistance index is calculated as 30 + 1 × 6 + 30 × 0.3 = 45 based on a motion trajectory deviation of 6 pulses and a phase current increment of 0.3A. All of these fall within their respective resistance index ranges.

[0104] First, the equipment status feedback value output from the preceding process is obtained. This feedback value characterizes the operating resistance index of the protective mechanism; the smaller the value, the smoother the mechanical action. A preset feedback threshold of 60 for the resistance index is set, based on the statistical upper limit of the resistance when the protective mechanism is normally deployed. This threshold is determined by statistically sampling the resistance index of multiple total stations performing normal deployment operations under both normal and low-temperature conditions during calibration tests of the protective cover's transmission mechanism. The sample mean of the resistance index under normal deployment is 42, and the sample standard deviation is 6. The statistical upper limit is calculated as 42 + 3 × 6 = 60, meaning the probability of the resistance index exceeding 60 under normal deployment is less than 0.15%. Therefore, 60 is used as the preset feedback threshold, and the protective cover is considered to be normally deployed only when the feedback value is below 60. For example, if the received device status feedback value is 72, which is not lower than the preset feedback threshold of 60, the timing matching is determined to be a failure. The position encoder signal and motor phase current data of the next 100ms time window are collected again and re-parsed. If the device status feedback value obtained after re-parse is 45, which is lower than the preset feedback threshold of 60, the main control unit generates a specific binary timing matching identifier and records the generation time of the protection activation command for subsequent timing verification.

[0105] Secondly, the time window for the channel blocking signal is determined based on the wake-up rhythm timing table pre-stored in the configuration storage area and the timing matching identifier. The wake-up rhythm timing table is pre-constructed based on the statistical beats of each stage in the total station wake-up process, including the expected duration of each of the four stages: power supply stabilization stage, initial excitation stage, self-test operation stage, and measurement preparation stage, which for example correspond to 10s, 15s, 35s, and 30s. The four stages are arranged continuously without gaps, with the time of remote wake-up command issuance as the unified time origin. The cumulative time boundaries are 0 to 10s, 10 to 25s, 25 to 60s, and 60 to 90s, respectively. The expected duration of the self-test operation stage is set according to the statistical beat of self-test under normal temperature conditions. The self-test time is extended under low temperature conditions, and the upper limit of the monitoring duration is set separately according to the statistical upper limit of low temperature. The completion time of the mechanical action is determined by monitoring the cumulative pulse count of the position encoder. The full extension stroke of the protective cover is 2mm. According to the conversion factor of 512 pulses per millimeter, the cumulative pulse count is 1024. When the cumulative pulse count reaches 1024, this moment is recorded as the completion time of the mechanical action.

[0106] The timing of the mechanical action is verified by checking whether the completion time falls within the expected time window (60 to 90 seconds) corresponding to the measurement preparation stage, based on the generation time of the protection activation command. For example, if the protection activation command is generated at 60 seconds, and the mechanical action unfolds a 2mm stroke at a speed of 5mm / s in 400ms (0.4s), the completion time of the mechanical action is 60.4s, falling within the 60 to 90 seconds range, confirming successful timing matching. If the completion time of the mechanical action is earlier than 60 seconds or later than 90 seconds, the timing matching is determined to be unsuccessful. The control motor is reset, and the mechanical action is re-executed, and the completion time is re-verified. The completion time after re-execution is still measured with the remote wake-up command issuance time as the unified time origin. Re-execution only changes the completion time itself and does not regenerate the time window. If the completion time after re-execution is earlier than 60 seconds, the control motor is decelerated. If the deployment speed is low, the process is re-executed to ensure the completion time falls within the time window. If the completion time after re-execution is later than 90 seconds, the timing correction within the current startup process is deemed irreversible because the completion time only increases over time and cannot be reversed by re-execution. In this case, there is no need to continue accumulating failure counts; it is directly considered as the equivalent final state that has reached the preset retry limit. The fault event of the protection mechanism is reported through the remote communication link, and the mechanical action is suspended, terminating the current startup process. The next startup process will start after the remote monitoring center reissues the remote wake-up command. The next startup process will re-measure the time window with the new remote wake-up command issuance time as the unified time origin. In other cases, if the accumulated failure count reaches the preset retry limit and still fails, the fault event of the protection mechanism is reported through the remote communication link, and the mechanical action is suspended.

[0107] Then, after successful timing matching confirmation, the main control unit immediately sends a continuous high-level channel blocking signal to the internal communication bus for 500ms. This duration is determined based on the statistical analysis of the total time consumed by the blocking signal coverage mechanism switching process, ensuring that the mechanism switching is completed within the duration of the blocking signal. This transmission time is still within the time window corresponding to the measurement preparation phase. When the bus receives the channel blocking signal, it immediately terminates the blind transmission mechanism of the underlying controller issuing default probe commands at a fixed frequency. This blind transmission mechanism is linked with the timed retransmission mechanism at the remote end. The channel blocking signal is synchronously reported as a start-up completion acknowledgment via the remote communication link, and the remote monitoring center stops command retransmission accordingly. Simultaneously, it stops outputting control current to the motor drive module and writes a specific activation instruction code to the blocking status register at memory address 10, completing the register's state flip.

[0108] If the device status feedback value is not lower than the preset feedback threshold, the timing match is determined to be unsuccessful. The position encoder signal and motor phase current data of the next 100ms time window are re-acquired. The device status feedback value is re-analyzed based on the newly acquired signal. If the device still fails after three retry attempts, the protective mechanism fault event is reported through the remote communication link and the mechanical action is suspended. The device waits for the remote monitoring center to initiate the manual intervention process. The preset retry limit is set at three times. The basis for this setting is that occasional jamming of the protective cover transmission mechanism is usually caused by instantaneous temperature deformation or small foreign objects. A single re-acquisition and analysis of the data in a 100ms window can eliminate the occasional factor. However, if three consecutive 100ms windows (a total of 300ms) are determined to be timing match failures, it indicates that there is a persistent mechanical fault in the transmission mechanism rather than occasional interference. Continuing to retry will not recover the device. Therefore, three times is set as the limit.

[0109] It should be noted that the verification of the equipment status feedback value is performed before the verification of the completion time of the mechanical action. When abnormal motion resistance and completion time exceeding the window occur simultaneously in the same round of verification, the abnormal motion resistance is handled first, followed by the completion time exceeding the window. That is, the system first returns to re-acquire and parse the position encoder signal to eliminate the abnormal motion resistance. After the equipment status feedback value is lower than the preset feedback threshold, the system then verifies whether the completion time of the mechanical action falls within the time window. If the completion time still does not fall within the time window at this time, the system controls the motor to reset and re-execute the mechanical action and re-verify the completion time. In other words, the handling of abnormal motion resistance has a higher priority than the handling of completion time exceeding the window, to avoid the simultaneous triggering of the two handling actions.

[0110] Finally, the communication handshake protocol parameters stored in the activated blocking status register are read, and the complete serial configuration parameters, such as baud rate 9600, 8 data bits, no parity bit, and 1 stop bit, are extracted. Based on the extracted protocol configuration data, the serial data transmission link between the main control unit and the optical measurement module is reinitialized. After the link configuration is completed, the angle and distance measurement modules inside the total station enter the power-on standby mode, establish a measurement preparation state, and wait to receive subsequent absolute position measurement commands.

[0111] In summary, this invention discloses a remote start method for automated monitoring of the electric protective cover of a total station. By collecting the power supply current during the total station's wake-up process, and after outlier removal and feature fusion, a support vector machine model identifies the operational phase and verifies the self-test status. A protective cover activation command is generated to control the motor to execute mechanical actions. The completion time of the mechanical actions is time-matched with the time window corresponding to the total station's wake-up rhythm. Upon successful matching, the blind start mechanism is terminated and a measurement preparation state is established. This invention solves the problem in existing technologies where the start-up timing of the protective cover is difficult to match with the total station's wake-up rhythm.

[0112] The second embodiment of the present invention provides a remote start system for the electric protective cover of a total station for automated monitoring, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0113] It should be noted that the remote start system for the electric protective cover of the total station automation monitoring provided in this embodiment of the invention is used to execute all the process steps of the remote start method for the electric protective cover of the total station automation monitoring described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0114] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0115] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for remotely starting the electric protective cover of an automated monitoring system for a total station, characterized in that, include: Obtain the remote wake-up command and wake up the total station, collect the power supply current during the wake-up process, and obtain the first current sequence; The first current sequence is subjected to outlier removal processing to remove electromagnetic interference spike data, resulting in the second current sequence; Based on the second current sequence, the time-domain current amplitude difference and frequency-domain energy distribution features are extracted and fused into a multi-dimensional feature vector; The multidimensional feature vector is input into a pre-trained support vector machine model for classification processing. The preset feature-stage mapping relationship is compared to determine the initial operation stage of the total station. When the initial operation stage is the uniform rotation stage, the internal status register is read to verify the self-test status. If it is the self-test completion stage, a protection activation command is generated. When the initial operation stage is the initial excitation stage or the self-test status is the incomplete self-test stage, the power supply current is collected and the multidimensional feature vector is updated. According to the protection activation command, the motor is controlled to perform mechanical actions. The position encoder signal during the operation of the motor is collected and analyzed to obtain the equipment status feedback value. If the equipment status feedback value is lower than the preset feedback threshold and the completion time of the mechanical action falls within the time window corresponding to the total station wake-up rhythm, it is determined that the timing of the mechanical action and the total station wake-up rhythm is successfully matched. The mechanism of the underlying controller periodically issuing the default detection command according to the blind transmission mechanism is terminated, and the measurement preparation state is established.

2. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The process of acquiring a remote wake-up command and waking up the total station, collecting the power supply current during the wake-up process, and obtaining a first current sequence includes: According to the remote wake-up command, the power supply circuit of the total station is triggered to close; Acquire the initial voltage signal of the power supply circuit. If the initial voltage signal is greater than a preset voltage threshold, collect the power supply current of the total station during the transition from sleep state to active state. The supply current is continuously sampled to obtain the first current sequence.

3. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The step of performing outlier removal processing on the first current sequence to remove electromagnetic interference spikes and obtain the second current sequence includes: The first current sequence is segmented by a sliding window to construct multidimensional feature space data; The average path length of the multidimensional feature space data is calculated using the isolated forest algorithm. Anomaly scores are calculated based on the average path length. Data segments with anomaly scores greater than a preset anomaly score threshold are identified as electromagnetic interference spike data. The electromagnetic interference spike data in the first current sequence is removed, and the sequence after removal is interpolated, filled, and smoothed to obtain the second current sequence.

4. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The step of extracting the time-domain current amplitude difference and frequency-domain energy distribution features based on the second current sequence and fusing them into a multi-dimensional feature vector includes: Calculate the current amplitude difference between adjacent sampling points in the second current sequence to obtain the time-domain current amplitude difference sequence; The second current sequence is subjected to Fourier transform to obtain a spectrum sequence. The spectrum sequence is divided into multiple frequency band intervals, and the energy proportion in each frequency band interval is calculated to obtain the frequency domain energy distribution characteristics. The time-domain current amplitude difference sequence is subjected to dimensionality reduction processing to obtain dimensionality reduction features, which are then concatenated with the frequency-domain energy distribution features to obtain fluctuation features; The fluctuation feature is filled with zero values ​​up to a preset dimension threshold to obtain the multidimensional feature vector.

5. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The step of inputting the multidimensional feature vectors into a pre-trained support vector machine model for classification processing, comparing the preset feature-stage mapping relationship, and determining the initial operating stage of the total station includes: The multidimensional feature vectors are input into a support vector machine model for mapping to obtain mapped feature data. Calculate the signed decision function value corresponding to the mapped feature data to obtain the decision value, perform a signed function operation on the decision value, and output the classification label; The stage identifier corresponding to the classification label is found in the preset feature-stage mapping relationship, and the initial operation stage is determined according to the stage identifier.

6. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The verification of self-check status includes: The running stage code in the internal status register is compared with a preset status mapping table to obtain a self-test status identifier, which includes the self-test completion stage and the self-test incomplete stage. When the self-test status is marked as the self-test completion stage, the current external ambient temperature data and wind speed data of the total station are extracted, and the external ambient temperature data and wind speed data are input into a preset instruction encoding rule library to generate the protection activation instruction containing environmental parameters. When the self-test status is marked as the incomplete self-test stage, the system returns to collect the power supply current and update the multidimensional feature vector. If the self-test completion stage is still not found after the preset monitoring time limit is reached, the system reports a self-test timeout event through the remote communication link and terminates the current startup process.

7. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 6, characterized in that, The updating of the multidimensional feature vector includes: Obtain the real-time power supply current data of the total station and extract the current data sequence within a preset current extraction window; The current data sequence is subjected to frequency domain analysis, and the amplitude at a preset frequency in the frequency domain signal is extracted as the current fluctuation feature. The current fluctuation feature is replaced with the feature component at a preset position in the multidimensional feature vector to complete the update of the multidimensional feature vector.

8. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The process of collecting and analyzing the position encoder signal during motor operation to obtain equipment status feedback values ​​includes: Edge detection is performed on the position encoder signal to obtain pulse sequence features; The angular displacement increment of the motor is determined based on the pulse sequence characteristics, and the difference between the angular displacement increment and the theoretical angular displacement increment is compared to obtain the motion trajectory deviation. If the deviation of the motion trajectory is greater than the preset deviation threshold, the action execution state corresponding to the motion trajectory deviation is analyzed, and the action execution state is mapped to a numerical resistance index according to the pre-stored state and value comparison relationship to obtain the device status feedback value. If the deviation of the motion trajectory is not greater than the preset deviation threshold, the resistance index corresponding to the normal execution state is used as the device status feedback value.

9. The remote start method for the electric protective cover of the total station's automated monitoring system according to claim 1, characterized in that, The determination that the timing of the mechanical action matches the total station's wake-up rhythm is successful, terminates the mechanism of the underlying controller periodically issuing default detection commands according to the blind firing mechanism, and establishes a measurement preparation state, including: When the device status feedback value is lower than the preset feedback threshold, a timing matching identifier is generated, and the generation time of the protection activation command is recorded. Based on the pre-stored wake-up rhythm timing table and the timing matching identifier, the transmission time window of the channel blocking signal is determined to be the time window corresponding to the total station wake-up rhythm. Combined with the generation time, the completion time of the mechanical action is verified to fall within the time window. Within the time window, the channel blocking signal is sent, and the mechanism for the underlying controller to periodically issue default probe commands according to the blind transmission mechanism is terminated based on the channel blocking signal, and the blocking status register is activated. Configure the data transmission link according to the communication handshake protocol in the blocking status register, and establish the measurement preparation state; If the completion time of the mechanical action does not fall within the time window, the timing match is determined to be unsuccessful. The motor is then reset and the mechanical action is re-executed and the completion time is re-verified. If the preset retry limit is reached and the failure still occurs, a fault event of the protection mechanism is reported through the remote communication link and the mechanical action is suspended.

10. A remote start system for an automated monitoring electric protective cover of a total station, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method according to any one of claims 1 to 9.

Images