Sensor data collection processing method and device, electronic equipment and medium

Abstract

The application relates to the technical field of sensor data processing, and discloses a sensor data collection and processing method and device, electronic equipment and a medium. The sensor data collection and processing method comprises the following steps: analyzing and processing sensor data of each sensor terminal to obtain a multi-dimensional difference feature vector corresponding to the sensor data of each sensor terminal; calculating the change strength of the sensor data according to the multi-dimensional difference feature vector; constructing a sampling parameter correlation function; configuring actual sampling parameters to the sensor terminal based on actual resource information of the sensor terminal and the sampling parameter correlation function; and updating the actual sampling parameters of the sensor terminal according to the actual change strength of the actual sensor data in a preset time window. According to the sensor data collection and processing method, the sampling parameters can be dynamically adjusted according to the actual change strength of the actual sensor data, and the adaptability of the sensor data collection strategy is improved.

Description

Technical Field

[0001] This application relates to the field of sensor data processing technology, and in particular to a method, apparatus, electronic device and medium for acquiring and processing sensor data. Background Technology

[0002] With the rapid development of intelligent manufacturing in industry, the factory of the future is evolving towards flexibility and intelligence. Massive numbers of sensor terminals are deployed within factories to collect multimodal data in real time during the production process, including equipment status data, environmental monitoring data, and product quality data. This data exhibits multi-source heterogeneity, meaning it comes from different manufacturers and different types of equipment, with significant differences in data format and sampling frequency. In related technologies, all sensor terminals typically employ a uniform, fixed sampling frequency and accuracy acquisition strategy. For slowly changing static data, excessively high sampling frequencies lead to resource waste; for rapidly changing dynamic data, fixed acquisition strategies struggle to capture data mutations, resulting in the loss of critical information. Summary of the Invention

[0003] This application provides a sensor data acquisition and processing method, apparatus, electronic device, and medium that can dynamically adjust the sampling frequency and sampling accuracy according to the actual change intensity of the actual sensor data. Furthermore, when configuring the actual sampling parameters, the actual resource information of the sensor terminal is associated, effectively balancing data quality and resource consumption, and improving the adaptability of the sensor data acquisition strategy.

[0004] In a first aspect, embodiments of this application provide a sensor data acquisition and processing method, comprising: analyzing and processing sensor data of each sensor terminal to obtain temporal difference features, frequency difference features, and spatial difference features of the sensor data corresponding to each sensor terminal; obtaining a multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the temporal difference features, the frequency difference features, and the spatial difference features; calculating the change intensity of the sensor data according to the multidimensional difference feature vector; constructing a sampling parameter association function, the sampling parameter association function including a mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal, the sampling parameters including sampling frequency and sampling accuracy; configuring actual sampling parameters to the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter association function, such that the sensor terminal acquires data using the actual sampling parameters to obtain actual sensor data; updating the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data according to a preset time window, wherein if the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased, and if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased.

[0005] According to some of the foregoing embodiments of the first aspect of this application, the step of analyzing and processing the sensor data of each sensor terminal to obtain the time-domain difference characteristics, frequency-domain difference characteristics, and spatial-domain difference characteristics of the sensor data corresponding to each sensor terminal includes: calculating the difference between adjacent sampling points of the sensor data in the time domain to obtain the time-domain difference characteristics; performing a fast Fourier transform on the sensor data to obtain the spectrum, extracting spectral features from the spectrum and processing them to obtain the frequency-domain difference characteristics; and calculating the spatial correlation parameters of the sensor data of adjacent sensor terminals in the spatial dimension to obtain the spatial-domain difference characteristics.

[0006] According to some of the foregoing embodiments of the first aspect of this application, the step of calculating the change intensity of the sensor data based on the multidimensional difference feature vector includes: processing the multidimensional difference feature vector using a normalization function to obtain the change intensity of the value in the range of 0 to 1.

[0007] According to some embodiments of the first aspect of this application, the resource information of the sensor terminal includes storage resource utilization, network bandwidth utilization, and computing resource utilization; the construction of the sampling parameter correlation function includes: constructing the sampling parameter correlation function according to the following formula: f_sample=f_base×(1+α·I_change)×(S_used / S_total)^(-β_s)×(B_used / B_total)^(-β_b)×(C_used / C_total)^(-β_c); p_sample=p_base×(1+γ·I_change)×(S_used / S_total)^(-δ_s)×(B_used / B_total)^(-δ_b); Where f_sample is the sampling frequency, p_sample is the sampling precision, f_base is the baseline sampling frequency, p_base is the baseline sampling precision, S_used / S_total is the storage resource utilization, I_change is the change intensity, B_used / B_total is the network bandwidth utilization, C_used / C_total is the computing resource utilization, α is the influence coefficient of change intensity on sampling frequency, γ is the influence coefficient of change intensity on sampling precision, β_s is the constraint coefficient of storage resource utilization on sampling frequency, β_b is the constraint coefficient of network bandwidth utilization on sampling frequency, β_c is the constraint coefficient of computing resource utilization on sampling frequency, δ_s is the constraint coefficient of storage resource utilization on sampling precision, and δ_b is the resource constraint coefficient of network bandwidth utilization on sampling precision.

[0008] According to some of the aforementioned embodiments of the first aspect of this application, the construction of the sampling parameter association function further includes: using a reinforcement learning model to optimize the data change influence coefficient and resource constraint coefficient in the sampling parameter association function.

[0009] According to some of the foregoing embodiments of the first aspect of this application, the sensor data acquisition and processing method further includes: establishing an acquisition quality evaluation model to acquire the sensor data for the period to be evaluated; acquiring a data integrity score, an accuracy score, and a timeliness score of the sensor data for the period to be evaluated through the acquisition quality evaluation model; and obtaining a comprehensive acquisition quality score corresponding to the sensor data for the period to be evaluated based on the data integrity score, the accuracy score, and the timeliness score.

[0010] According to some of the foregoing embodiments of the first aspect of this application, the sensor data acquisition and processing method further includes: acquiring task metadata of the task corresponding to the actual sensor data; identifying the data set that the task depends on, and acquiring the importance weight of each data in the data set to the task, thereby obtaining an importance weight set; obtaining the actual data value of the task based on the task metadata, the importance weight set, and a preset data value model; and determining whether to perform error correction processing on the actual sensor data based on the actual data value.

[0011] According to some of the foregoing embodiments of the first aspect of this application, the sensor data acquisition and processing method further includes: after determining that the actual sensor data will be corrected, performing abnormal data detection on the actual sensor data, removing abnormal data points from the actual sensor data, and obtaining cleaned sensor data; and interpolating the abnormal data points and missing data points in the actual sensor data based on the cleaned sensor data to obtain corrected sensor data.

[0012] According to some of the foregoing embodiments of the first aspect of this application, the step of detecting abnormal data in the actual sensor data, removing abnormal data points from the actual sensor data, and obtaining cleaned sensor data includes: encoding and compressing the actual sensor data using a bidirectional long short-term memory network encoder to obtain compressed data; reconstructing the compressed data using a bidirectional long short-term memory network decoder to obtain reconstructed data; calculating the reconstruction error between the reconstructed data and the actual sensor data; marking data points with reconstruction errors greater than a preset abnormal threshold as abnormal data points; and filtering out the abnormal data points from the actual sensor data to obtain cleaned sensor data.

[0013] According to some of the foregoing embodiments of the first aspect of this application, the step of interpolating abnormal data points and missing data points in the actual sensor data based on the cleaned sensor data to obtain error-corrected sensor data includes: extracting spatial correlation features of the sensor terminal using a multi-layer convolutional neural network model; extracting temporal correlation features of the cleaned sensor data using a multi-layer long short-term memory network model; constructing a nonlinear interpolation function based on the cleaned sensor data, the spatial correlation features, and the temporal correlation features; and interpolating abnormal data points and missing data points in the actual sensor data based on the nonlinear interpolation function to obtain error-corrected sensor data.

[0014] Secondly, embodiments of this application provide a sensor data acquisition and processing device, comprising: a multidimensional difference feature vector acquisition module, used to analyze and process sensor data of each sensor terminal, obtain time-domain difference features, frequency-domain difference features, and spatial-domain difference features of the sensor data corresponding to each sensor terminal, and obtain a multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the time-domain difference features, the frequency-domain difference features, and the spatial-domain difference features; a change intensity acquisition module, used to calculate the change intensity of the sensor data according to the multidimensional difference feature vector; and an association function construction module, used to construct a sampling parameter association function, wherein the sampling parameter association function includes the sampling parameters of the sensor terminal and the change intensity. The system establishes a mapping relationship between intensity and the resource information of the sensor terminal. The sampling parameters include sampling frequency and sampling accuracy. An actual sampling parameter configuration module is used to configure actual sampling parameters for the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter association function, so that the sensor terminal collects data using the actual sampling parameters to obtain actual sensor data. A dynamic adjustment module is used to update the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data within a preset time window. If the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased.

[0015] According to some embodiments of the second aspect of this application, the sensor data acquisition and processing device further includes: a task element extraction module, used to acquire task element information of the task corresponding to the actual sensor data; a data dependency analysis module, used to identify the data set that the task depends on, and acquire the importance weight of each data in the data set to the task, to obtain an importance weight set; a data value modeling module, used to obtain the actual data value of the task based on the task element information, the importance weight set, and a preset data value model; an error correction decision module, used to determine whether to perform error correction processing on the actual sensor data based on the actual data value; an anomaly detection module, used to perform anomaly data detection on the actual sensor data after determining that error correction processing should be performed on the actual sensor data, remove abnormal data points in the actual sensor data, and obtain cleaned sensor data; and an error correction module, used to interpolate the abnormal data points and missing data points in the actual sensor data based on the cleaned sensor data, to obtain error-corrected sensor data.

[0016] Thirdly, embodiments of this application provide an electronic device, which includes a memory and a processor. The memory is communicatively connected to the processor. The memory stores instructions, and the processor calls the instructions in the memory to cause the electronic device to execute a sensor data acquisition and processing method according to any of the foregoing embodiments of the first aspect of this application.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing instructions that, when executed by a processor, implement a sensor data acquisition and processing method according to any of the foregoing embodiments of the first aspect of this application.

[0018] The sensor data acquisition and processing method according to an embodiment of this application includes: analyzing and processing sensor data from each sensor terminal to obtain temporal, frequency, and spatial difference features of the sensor data corresponding to each sensor terminal; obtaining a multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the temporal, frequency, and spatial difference features; calculating the change intensity of the sensor data based on the multidimensional difference feature vector; constructing a sampling parameter association function, which includes a mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal, wherein the sampling parameters include sampling frequency and sampling accuracy; configuring actual sampling parameters to the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter association function; updating the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data within a preset time window, wherein if the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased. In the above scheme, by constructing a sampling parameter association function, dynamic adaptive adjustment of the sampling frequency and sampling accuracy is achieved. Specifically, the sampling frequency and accuracy are dynamically adjusted based on the actual intensity of changes in the sensor data. Sampling parameters are automatically increased when data changes drastically and decreased when data is stable. Furthermore, the actual sampling parameters are configured in conjunction with the actual resource information of the sensor terminal, effectively balancing data quality and resource consumption, and enhancing the adaptability of the sensor data acquisition strategy. In this scheme, differentiated actual sampling parameters are configured for the sensor terminal based on its actual resource information and the correlation function of sampling parameters. This allows for the customization of suitable acquisition strategies for different sensor terminals, significantly improving the acquisition efficiency and quality of multi-source heterogeneous sensor data, and achieving high-quality acquisition and processing of multi-source heterogeneous sensor data under resource-constrained conditions. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0020] Figure 1 This is a flowchart of some steps of an embodiment of the sensor data acquisition and processing method according to this application; Figure 2 This is a flowchart of another part of the steps of one embodiment of the sensor data acquisition and processing method according to this application; Figure 3 This is a flowchart of another part of the steps of an embodiment of the sensor data acquisition and processing method according to this application; Figure 4This is a flowchart illustrating how, in one embodiment of the sensor data acquisition and processing method of this application, abnormal data detection is performed on actual sensor data, abnormal data points are removed from the actual sensor data, and cleaned sensor data is obtained. Figure 5 This is a flowchart illustrating how, in one embodiment of the sensor data acquisition and processing method of this application, abnormal data points and missing data points in the actual sensor data are interpolated based on the cleaned sensor data to obtain the error-corrected sensor data. Figure 6 This is a schematic diagram of the structure of one embodiment of the sensor data acquisition and processing device according to this application; Figure 7 This is a schematic diagram of the hardware structure of an embodiment of the electronic device according to this application. Detailed Implementation

[0021] The technical solutions in the embodiments (or "implementations") of this application will be clearly and completely described herein with reference to the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements.

[0022] If the embodiments of this application contain terms relating to directional indications or positional relationships (such as up, down, left, right, front, back, inside, outside, top, bottom, center, vertical, horizontal, longitudinal, transverse, length, width, counterclockwise, clockwise, axial, radial, circumferential, etc.), such terms are only used to explain the relative positional relationships and movement of the components in a specific posture (as shown in the attached figures); if the specific posture changes, the directional indications or positional relationships will also change accordingly. Furthermore, the terms "first" and "second" used in the embodiments of this application are only for descriptive convenience and should not be construed as indicating or implying relative importance.

[0023] With the rapid development of intelligent manufacturing in industry, the factory of the future is evolving towards flexibility and intelligence. Massive numbers of sensor terminals are deployed within the factory to collect multimodal data in real time during the production process, including equipment status data, environmental monitoring data, and product quality data. This data exhibits multi-source heterogeneity, meaning it comes from different manufacturers and different types of equipment, with significant differences in data format and sampling frequency.

[0024] In related technologies, all sensor terminals typically adopt a uniform and fixed sampling frequency and accuracy acquisition strategy. For slowly changing static data, an excessively high sampling frequency leads to a waste of resources; for rapidly changing dynamic data, a fixed acquisition strategy is difficult to capture data mutations, resulting in the loss of key information.

[0025] In some scenarios, sensor terminals have limited resources, making it difficult to support high-frequency, high-precision acquisition of all data. Related technologies, under resource-constrained conditions, struggle to guarantee optimal sensor data acquisition quality.

[0026] This application provides a method for acquiring and processing sensor data. The model is an artificial intelligence model. Figure 1 This is a flowchart of some steps of an embodiment of the sensor data acquisition and processing method according to this application. The sensor data acquisition and processing method includes steps S110 to S113 and steps S121 to S122.

[0027] In step S111, the sensor data of each sensor terminal is analyzed and processed to obtain the time-domain difference characteristics, frequency-domain difference characteristics and spatial-domain difference characteristics of the sensor data corresponding to each sensor terminal. Based on the time-domain difference characteristics, frequency-domain difference characteristics and spatial-domain difference characteristics, a multi-dimensional difference feature vector of the sensor data corresponding to each sensor terminal is obtained.

[0028] In some embodiments, the types of sensor data from the sensor terminal include time-series data, image data, text data, etc. The following explanation will use the example of time-series data from the sensor terminal, such as sensor data showing continuous changes in temperature, vibration, or pressure over time.

[0029] In some embodiments, the sensor data of each sensor terminal is analyzed and processed to obtain the temporal difference characteristics, frequency difference characteristics, and spatial difference characteristics of the sensor data corresponding to each sensor terminal. This includes: calculating the difference between adjacent sampling points of the sensor data in the time domain to obtain the temporal difference characteristics; performing a Fast Fourier Transform (FFT) on the sensor data to obtain the spectrum, extracting spectral features from the spectrum and processing them to obtain the frequency difference characteristics; and calculating the spatial correlation parameters of the sensor data of adjacent sensor terminals in the spatial dimension to obtain the spatial difference characteristics.

[0030] In one example, the difference between adjacent sampling points of sensor data in the time domain is calculated as D_t=|x(t)-x(t-1)|, which analyzes the rate of change of sensor data in the time dimension and obtains the time domain difference characteristics. The time series data of the sensor data is subjected to Fast Fourier Transform to obtain the spectrum. The spectral features are extracted from the spectrum and processed, which analyzes the differences of sensor data in the frequency dimension and obtains the frequency domain difference characteristics. For a spatially distributed sensor terminal network, the spatial correlation of data of adjacent sensor terminal nodes is calculated to obtain the spatial domain difference characteristics.

[0031] In some embodiments, obtaining a multidimensional difference feature vector for the sensor data corresponding to each sensor terminal based on time-domain difference features, frequency-domain difference features, and spatial-domain difference features includes: fusing the time-domain difference features, frequency-domain difference features, and spatial-domain difference features into a multidimensional difference feature vector V_diff, which characterizes the comprehensive characteristics of the dynamic changes in sensor data. In one example, fusing the time-domain difference features, frequency-domain difference features, and spatial-domain difference features into a multidimensional difference feature vector V_diff involves directly assembling the time-domain difference features, frequency-domain difference features, and spatial-domain difference features into a vector to obtain the multidimensional difference feature vector V_diff.

[0032] In step S112, the intensity of change in the sensor data is calculated based on the multidimensional difference feature vector.

[0033] In some embodiments, calculating the change intensity of sensor data based on a multidimensional difference feature vector includes: processing the multidimensional difference feature vector using a normalization function to obtain the change intensity with values ​​in the range of 0 to 1.

[0034] In one example, the intensity of change in sensor data, I_change=f(V_diff), is calculated based on the multidimensional difference feature vector V_diff, where f(V_diff) is a normalization function, thereby mapping the intensity of change I_change to the interval [0, 1].

[0035] In step S113, a sampling parameter correlation function is constructed. The sampling parameter correlation function includes the mapping relationship between the sampling parameters of the sensor terminal and the intensity of change and the resource information of the sensor terminal. The sampling parameters include sampling frequency and sampling accuracy.

[0036] In some embodiments, the resource information of the sensor terminal includes storage resource utilization S_used / S_total, network bandwidth utilization B_used / B_total, and computing resource utilization C_used / C_total; constructing the sampling parameter correlation function includes: constructing the sampling parameter correlation function according to the following formula: f_sample=f_base×(1+α·I_change)×(S_used / S_total)^(-β_s)×(B_used / B_total)^(-β_b)×(C_used / C_total)^(-β_c); p_sample=p_base×(1+γ·I_change)×(S_used / S_total)^(-δ_s)×(B_used / B_total)^(-δ_b); Wherein, f_sample is the sampling frequency, p_sample is the sampling precision, f_base is the baseline sampling frequency, p_base is the baseline sampling precision, S_used / S_total is the storage resource utilization, I_change is the change intensity, B_used / B_total is the network bandwidth utilization, C_used / C_total is the computational resource utilization, α and γ are the change intensity influence coefficients, and β_s, β_b, β_c, δ_s, and δ_b are resource constraint coefficients. Specifically, α is the influence coefficient of change intensity on sampling frequency, γ is the influence coefficient of change intensity on sampling precision, β_s is the constraint coefficient of storage resource utilization on sampling frequency, β_b is the constraint coefficient of network bandwidth utilization on sampling frequency, β_c is the constraint coefficient of computational resource utilization on sampling frequency, δ_s is the constraint coefficient of storage resource utilization on sampling precision, and δ_b is the resource constraint coefficient of network bandwidth utilization on sampling precision. The sampling parameters include the sampling frequency f_sample and the sampling precision p_sample. The data change influence coefficient and resource constraint coefficient can be obtained through training with historical data.

[0037] In some embodiments, constructing the sampling parameter association function further includes: using a reinforcement learning model to optimize the data change influence coefficient and resource constraint coefficient in the sampling parameter association function. In some embodiments, the reinforcement learning model is used to optimize the data change influence coefficient and resource constraint coefficient in the sampling parameter association function, and the optimization objective is to maximize the acquisition quality Q=Σ_i(w_i·accuracy_i), where w_i is the value weight of sensor data i, and accuracy_i is the acquisition accuracy of sensor data i.

[0038] In step S121, based on the actual resource information of the sensor terminal and the correlation function of the sampling parameters, the actual sampling parameters are configured to the sensor terminal, so that the sensor terminal can collect data with the actual sampling parameters to obtain actual sensor data.

[0039] In some embodiments, the actual resource information of each sensor terminal is monitored in real time. The actual resource information includes the actual storage resource utilization rate, the actual network bandwidth utilization rate, and the actual computing resource utilization rate.

[0040] In some embodiments, sensor terminals are classified according to their device type and capability differences. In some embodiments, a capability feature vector C_device = {C_storage, C_bandwidth, C_compute, C_power} is established for the sensor terminal, where C_storage represents storage resource characteristics, C_bandwidth represents network bandwidth characteristics, C_compute represents computing resource characteristics, and C_power represents power resource characteristics.

[0041] In some embodiments, based on the actual resource information of the sensor terminal and the sampling parameter association function, actual sampling parameters {f_sample_j, p_sample_j} are configured to the sensor terminal, where j represents the device number of the sensor terminal.

[0042] In step S122, the actual sampling parameters of the sensor terminal are updated according to the actual change intensity of the actual sensor data within a preset time window. If the actual change intensity is greater than the first threshold, the value of the actual sampling parameters is increased; if the actual change intensity is less than the second threshold, the value of the actual sampling parameters is decreased.

[0043] In one example, if the actual change intensity I_change is greater than the first threshold θ_high, the value of the actual sampling parameter is increased, that is, the actual sampling frequency and the actual sampling accuracy are increased; if the actual change intensity I_change is less than the second threshold θ_low, the value of the actual sampling parameter is decreased, that is, the actual sampling frequency and the actual sampling accuracy are decreased.

[0044] Figure 2 This is a flowchart of another part of the steps of one embodiment of the sensor data acquisition and processing method according to this application. In some embodiments, the sensor data acquisition and processing method further includes steps S131 to S133.

[0045] In step S131, a data acquisition quality evaluation model is established to obtain sensor data for the time period to be evaluated.

[0046] In step S132, the data integrity score, accuracy score, and timeliness score of the sensor data for the period to be evaluated are obtained by acquiring the quality evaluation model.

[0047] In step S133, a comprehensive quality score for the sensor data acquisition during the evaluation period is obtained based on the data integrity score, accuracy score, and timeliness score.

[0048] In one example, a data integrity score (C_completeness), an accuracy score (C_accuracy), and a timeliness score (C_timeliness) for sensor data during the evaluation period are obtained through a data acquisition quality assessment model. Based on these scores, a comprehensive acquisition quality score (Q_total) for the sensor data during the evaluation period is obtained: Q_total = λ_1·C_completeness + λ_2·C_accuracy + λ_3·C_timeliness, where λ_1 is the data integrity weight, λ_2 is the accuracy weight, and λ_3 is the timeliness weight.

[0049] Figure 3 This is a flowchart of another portion of the steps of an embodiment of the sensor data acquisition and processing method according to this application. In some embodiments, the sensor data acquisition and processing method further includes steps S141 to S144.

[0050] In step S141, the task metadata of the task corresponding to the actual sensor data is obtained.

[0051] In one example, the task metadata includes task type T_type, priority P_task, deadline T_deadline, and resource requirement R_require.

[0052] In step S142, the data set that the task depends on is identified, and the importance weight of each data in the data set to the task is obtained, thus obtaining the importance weight set.

[0053] In one example, the dataset D_dep = {d_1, d_2, ..., d_m} that the task depends on is identified. The importance weights of each data point in the dataset to the task are obtained, resulting in the importance weight set W_impact = {w_1, w_2, ..., w_m}.

[0054] In step S143, the actual data value of the task is obtained based on the task metadata, the set of importance weights, and the preset data value model.

[0055] In some embodiments, a data value model is constructed. This data value model is V_data(d) = f_value(T_type, P_task, W_impact, R_require, T_deadline). Based on task metadata, the set of importance weights, and this data value model, the actual data value of the task is obtained.

[0056] In step S144, it is determined whether to perform error correction processing on the actual sensor data based on the actual data value.

[0057] In one example, if the actual data value is greater than a first value threshold, it is determined that the actual sensor data should be corrected; if the actual data value is less than a second value threshold, it is determined that the actual sensor data should not be corrected.

[0058] In one example, if the actual data value V_data is greater than the first value threshold V_high, corresponding to high-value data, and resources are sufficient, it is determined to perform full error correction on the actual sensor data; if the actual data value V_data is greater than or equal to the second value threshold V_low and less than or equal to the first value threshold V_high, corresponding to medium-value data, it is determined to perform approximate error correction on the actual sensor data; if the actual data value V_data is less than the second value threshold V_low, corresponding to low-value data, it is determined not to perform error correction on the actual sensor data, skip the error correction step, and retain the original data.

[0059] like Figure 3 In some embodiments, the sensor data acquisition and processing method further includes steps S151 and S152.

[0060] In step S151, after determining that error correction processing should be performed on the actual sensor data, abnormal data detection is performed on the actual sensor data to remove abnormal data points in the actual sensor data, thereby obtaining cleaned sensor data.

[0061] Figure 4 This is a flowchart illustrating how, in one embodiment of the sensor data acquisition and processing method of this application, abnormal data is detected in actual sensor data, abnormal data points are removed from the actual sensor data, and cleaned sensor data is obtained. In some embodiments, the process of detecting abnormal data in actual sensor data, removing abnormal data points from the actual sensor data, and obtaining cleaned sensor data includes steps S1511 to S1515.

[0062] In some embodiments, prior to step S1511, abnormal data detection is performed on the actual sensor data to remove abnormal data points, resulting in cleaned sensor data. The method further includes dimensionality reduction processing of the actual sensor data. In one example, leveraging the sparsity and distance clustering effect of high-dimensional data, principal component analysis is used to reduce the dimensionality of the actual sensor data, retaining the main feature components and reducing computational complexity.

[0063] In step S1511, a bidirectional long short-term memory (Bi-LSTM) encoder is used to encode and compress the actual sensor data to obtain compressed data.

[0064] In one example, a bidirectional long short-term memory network encoder is used to encode and compress the actual sensor data. The time-series data of the actual sensor data is encoded and compressed, and the sequence X={x_1,x_2,...,x_L} of length L is encoded into a fixed-dimensional latent space representation, resulting in compressed data H=BiLSTM_encode(X).

[0065] In step S1512, the compressed data is reconstructed using a bidirectional long short-term memory network decoder to obtain the reconstructed data.

[0066] In one example, a bidirectional long short-term memory network decoder is used to reconstruct the compressed data H, resulting in reconstructed data X_prime=BiLSTM_decode(H), where X_prime={x_prime_1,x_prime_2,...,x_prime_L}.

[0067] In step S1513, the reconstruction error between the reconstructed data and the actual sensor data is calculated.

[0068] In one example, the reconstruction error E_rec=sqrt(Σ_i(x_i-x_prime_i)^2).

[0069] In step S1514, data points whose reconstruction error is greater than a preset abnormal threshold are marked as abnormal data points.

[0070] In some embodiments, the anomaly threshold can be set according to the reconstruction error distribution of historical normal data: anomaly threshold T_anomaly = μ_error + k·σ_error, where μ_error is the mean reconstruction error, σ_error is the standard deviation, and k is the safety factor (e.g., 3).

[0071] Data points whose reconstruction error E_rec is greater than the preset anomaly threshold T_anomaly are marked as anomaly data points.

[0072] In step S1515, abnormal data points are filtered out from the actual sensor data to obtain cleaned sensor data X_clean.

[0073] In step S152, based on the cleaned sensor data, the abnormal data points and missing data points in the actual sensor data are interpolated to obtain the error-corrected sensor data.

[0074] Figure 5This is a flowchart illustrating how, in one embodiment of the sensor data acquisition and processing method according to this application, abnormal data points and missing data points in actual sensor data are interpolated based on cleaned sensor data to obtain error-corrected sensor data. In some embodiments, interpolating abnormal data points and missing data points in actual sensor data based on cleaned sensor data to obtain error-corrected sensor data includes steps S1521 to S1524.

[0075] In step S1521, a multi-layer convolutional neural network model is used to extract the spatial correlation features F_spatial of the sensor terminal.

[0076] In step S1522, a multilayer long short-term memory network model is used to extract the temporal correlation features F_temporal of the cleaned sensor data.

[0077] In step S1523, a nonlinear interpolation function is constructed based on the cleaned sensor data, spatial correlation features, and temporal correlation features.

[0078] In one example, a nonlinear interpolation function x_corrected(t) = f_interp(F_spatial, F_temporal, X_clean) is constructed based on the cleaned sensor data X_clean, the spatial correlation feature F_spatial, and the temporal correlation feature F_temporal, where f_interp(F_spatial, F_temporal, X_clean) is an interpolation function constructed based on a neural network model.

[0079] In step S1524, based on a nonlinear interpolation function, abnormal data points and missing data points in the actual sensor data are interpolated to obtain the error-corrected sensor data.

[0080] In one example, for outlier data points, the interpolation results of the normal data points around the outlier data point are used to replace the data; for missing data points, the data of the time points before and after the missing data point and the spatially adjacent points are used for interpolation to complete the missing data point.

[0081] The sensor data acquisition and processing method according to an embodiment of this application includes: analyzing and processing sensor data from each sensor terminal to obtain temporal, frequency, and spatial difference features of the sensor data corresponding to each sensor terminal; obtaining a multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the temporal, frequency, and spatial difference features; calculating the change intensity of the sensor data based on the multidimensional difference feature vector; constructing a sampling parameter association function, which includes a mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal, wherein the sampling parameters include sampling frequency and sampling accuracy; configuring actual sampling parameters to the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter association function; updating the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data within a preset time window, wherein if the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased. In the above scheme, by constructing a sampling parameter association function, dynamic adaptive adjustment of the sampling frequency and sampling accuracy is achieved. Specifically, the sampling frequency and accuracy are dynamically adjusted based on the actual intensity of changes in the sensor data. Sampling parameters are automatically increased when data changes drastically and decreased when data is stable. Furthermore, the configuration of actual sampling parameters is linked to the actual resource information of the sensor terminal, effectively balancing data quality and resource consumption, and improving the adaptability of the sensor data acquisition strategy. Even in resource-constrained environments, optimal configuration of sensor data acquisition quality can be achieved, enabling high-quality acquisition and processing of multi-source heterogeneous sensor data under resource-limited conditions.

[0082] According to the sensor data acquisition and processing method of this application embodiment, based on the actual resource information of the sensor terminal and the sampling parameter correlation function, differentiated actual sampling parameters are configured for the sensor terminal, and suitable acquisition strategies are customized for different sensor terminals, which significantly improves the acquisition efficiency and quality of multi-source heterogeneous sensor data.

[0083] In some embodiments, the sensor data acquisition and processing method further includes: after determining that error correction processing should be performed on the actual sensor data, performing abnormal data detection on the actual sensor data, removing abnormal data points in the actual sensor data, and obtaining cleaned sensor data; based on the cleaned sensor data, interpolating the abnormal data points and missing data points in the actual sensor data to obtain error-corrected sensor data, thereby realizing abnormal detection, data correction, and missing data completion of the actual sensor data, and significantly improving data fidelity.

[0084] In some embodiments, anomaly detection is performed on actual sensor data to remove abnormal data points and obtain cleaned sensor data. This includes: encoding and compressing the actual sensor data using a bidirectional long short-term memory network encoder to obtain compressed data; reconstructing the compressed data using a bidirectional long short-term memory network decoder to obtain reconstructed data; calculating the reconstruction error between the reconstructed data and the actual sensor data; marking data points with reconstruction errors greater than a preset anomaly threshold as anomaly data points; and filtering out the anomaly data points from the actual sensor data to obtain cleaned sensor data. The above scheme uses a bidirectional long short-term memory network encoder for anomaly detection, which can effectively capture the long-term dependencies of the actual sensor data time series and has a higher detection accuracy compared to traditional threshold-based anomaly detection methods.

[0085] In some embodiments, the sensor data acquisition and processing method further includes: acquiring task metadata corresponding to the actual sensor data; identifying the data set that the task depends on, and obtaining the importance weight of each data point in the data set to the task, thus obtaining an importance weight set; obtaining the actual data value of the task based on the task metadata, the importance weight set, and a preset data value model; and determining whether to perform error correction processing on the actual sensor data based on the actual data value. In the above scheme, selective error correction is performed based on the actual data value of the task. When resources are limited, priority can be given to ensuring the error correction quality of high-value data, avoiding ineffective computational overhead for low-value data, and improving resource utilization efficiency.

[0086] In some embodiments, when computing resources are limited, priority is given to ensuring the error correction quality of high-value data, and optimal resource allocation is achieved by dynamically adjusting the number of iterations and precision parameters of the error correction algorithm.

[0087] In some embodiments, the sensor data acquisition and processing method further includes: establishing an acquisition quality evaluation model to obtain sensor data for the evaluation period; obtaining data integrity score, accuracy score, and timeliness score of the sensor data for the evaluation period through the acquisition quality evaluation model; and obtaining a comprehensive acquisition quality score corresponding to the sensor data for the evaluation period based on the data integrity score, accuracy score, and timeliness score. In the above scheme, an acquisition quality evaluation model based on multi-dimensional evaluation of data integrity, accuracy, and timeliness is established, providing quantitative indicators for system optimization. This comprehensive evaluation system can more comprehensively reflect the overall quality of data acquisition and provide more accurate guidance for parameter optimization.

[0088] This application also provides a sensor data acquisition and processing device. Figure 6This is a schematic diagram of one embodiment of the sensor data acquisition and processing device according to this application. The sensor data acquisition and processing device includes a multidimensional difference feature vector acquisition module 111, a change intensity acquisition module 112, a correlation function construction module 113, an actual sampling parameter configuration module 121, and a dynamic adjustment module 122.

[0089] The multidimensional difference feature vector acquisition module 111 is used to analyze and process the sensor data of each sensor terminal, obtain the time domain difference features, frequency domain difference features and spatial domain difference features of the sensor data corresponding to each sensor terminal, and obtain the multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the time domain difference features, frequency domain difference features and spatial domain difference features.

[0090] The change intensity acquisition module 112 is used to calculate the change intensity of sensor data based on the multidimensional difference feature vector.

[0091] The correlation function construction module 113 is used to construct the sampling parameter correlation function. The sampling parameter correlation function includes the mapping relationship between the sampling parameters of the sensor terminal and the intensity of change and the resource information of the sensor terminal. The sampling parameters include the sampling frequency and the sampling accuracy.

[0092] The actual sampling parameter configuration module 121 is used to configure the actual sampling parameters to the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter association function, so that the sensor terminal can collect data with the actual sampling parameters to obtain actual sensor data.

[0093] The dynamic adjustment module 122 is used to update the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data within a preset time window. If the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased.

[0094] According to the sensor data acquisition and processing apparatus of this application embodiment, the multidimensional difference feature vector acquisition module 111 is used to analyze and process the sensor data of each sensor terminal to obtain the time-domain difference features, frequency-domain difference features, and spatial-domain difference features of the sensor data corresponding to each sensor terminal, and to obtain the multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the time-domain difference features, frequency-domain difference features, and spatial-domain difference features. The change intensity acquisition module 112 is used to calculate the change intensity of the sensor data based on the multidimensional difference feature vector. The correlation function construction module 113 is used to construct a sampling parameter correlation function, which includes the mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal. The sampling parameters include the sampling frequency and the sampling accuracy. The actual sampling parameter configuration module 121 is used to configure the actual sampling parameters of the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter correlation function, so that the sensor terminal performs data acquisition with the actual sampling parameters to obtain actual sensor data. The dynamic adjustment module 122 updates the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data within a preset time window. Specifically, if the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased. In this scheme, a sampling parameter correlation function is constructed to achieve dynamic adaptive adjustment of the sampling frequency and sampling accuracy. Specifically, the sampling frequency and sampling accuracy are dynamically adjusted according to the actual change intensity of the actual sensor data. The sampling parameters are automatically increased when data changes drastically and decreased when data is stable. Furthermore, the actual sampling parameters are configured in conjunction with the actual resource information of the sensor terminal, effectively balancing data quality and resource consumption, and improving the adaptability of the sensor data acquisition strategy. Even in resource-constrained environments, optimal configuration of sensor data acquisition quality can be achieved, enabling high-quality acquisition and processing of multi-source heterogeneous sensor data under resource-constrained conditions.

[0095] According to the sensor data acquisition and processing apparatus of this application embodiment, based on the actual resource information of the sensor terminal and the sampling parameter correlation function, differentiated actual sampling parameters are configured for the sensor terminal, and suitable acquisition strategies are customized for different sensor terminals, which significantly improves the acquisition efficiency and quality of multi-source heterogeneous sensor data.

[0096] In some embodiments, the sensor data acquisition and processing device further includes a task element extraction module 141, a data dependency analysis module 142, a data value modeling module 143, and an error correction decision module 144.

[0097] The task element extraction module 141 is used to obtain task element information of the task corresponding to the actual sensor data.

[0098] The data dependency analysis module 142 is used to identify the data set that the task depends on, and to obtain the importance weight of each data in the data set to the task, thus obtaining the importance weight set.

[0099] The data value modeling module 143 is used to obtain the actual data value of a task based on task metadata, a set of importance weights, and a preset data value model.

[0100] The error correction decision module 144 is used to determine whether to perform error correction processing on the actual sensor data based on the actual data value.

[0101] In the above embodiments, selective error correction can be performed based on the actual data value of the task. When resources are limited, priority can be given to ensuring the error correction quality of high-value data, avoiding unnecessary computational overhead for low-value data, and improving resource utilization efficiency.

[0102] In some embodiments, the sensor data acquisition and processing device further includes an anomaly detection module 151 and an error correction module 152.

[0103] The anomaly detection module 151 is used to detect abnormal data in the actual sensor data after determining that error correction processing has been performed on the actual sensor data, remove abnormal data points in the actual sensor data, and obtain cleaned sensor data.

[0104] The error correction module 152 is used to interpolate abnormal and missing data points in the actual sensor data based on the cleaned sensor data to obtain the error-corrected sensor data.

[0105] In the above embodiments, the anomaly detection module 151 is used to detect anomalies in the actual sensor data after determining that error correction processing should be performed on the actual sensor data, remove abnormal data points in the actual sensor data, and obtain cleaned sensor data; the error correction module 152 is used to interpolate the abnormal data points and missing data points in the actual sensor data based on the cleaned sensor data to obtain error-corrected sensor data, thereby realizing anomaly detection, data correction, and missing data completion of the actual sensor data, significantly improving data fidelity.

[0106] This application also provides an electronic device. Figure 7 This is a schematic diagram of the hardware structure of an embodiment of the electronic device according to this application. The electronic device includes a memory 910 and a processor 920. The memory 910 and the processor 920 are communicatively connected. The memory 910 stores instructions, and the processor 920 calls the instructions in the memory 910 to cause the electronic device to execute the sensor data acquisition and processing method according to any of the foregoing embodiments of this application.

[0107] The sensor data acquisition and processing method includes: analyzing and processing the sensor data of each sensor terminal to obtain the temporal, frequency, and spatial difference characteristics of the sensor data corresponding to each sensor terminal; obtaining a multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the temporal, frequency, and spatial difference characteristics; calculating the change intensity of the sensor data based on the multidimensional difference feature vector; constructing a sampling parameter correlation function, which includes the mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal; the sampling parameters include sampling frequency and sampling accuracy; configuring actual sampling parameters for the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter correlation function, so that the sensor terminal can acquire data with the actual sampling parameters to obtain actual sensor data; updating the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data according to a preset time window, wherein if the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased, and if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased.

[0108] Specifically, the processor 920 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0109] Memory 910 may include a large-capacity memory for data or instructions. For example, and not limitingly, memory 910 may include a hard disk drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 910 may include removable or non-removable (or fixed) media. Where appropriate, memory 910 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 910 may be the non-volatile memory described above. In a particular embodiment, memory 910 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0110] In one example, the electronic device may also include a communication interface 930 and a bus 940. The processor 920, memory 910, and communication interface 930 are connected via the bus 940 and communicate with each other.

[0111] The communication interface 930 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0112] Bus 940 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 940 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0113] Furthermore, in conjunction with the sensor data acquisition and processing methods in the above embodiments, this application can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores instructions that, when executed by a processor, implement the sensor data acquisition and processing methods of any of the foregoing embodiments of this application.

[0114] The sensor data acquisition and processing method includes: analyzing and processing the sensor data of each sensor terminal to obtain the temporal, frequency, and spatial difference characteristics of the sensor data corresponding to each sensor terminal; obtaining a multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the temporal, frequency, and spatial difference characteristics; calculating the change intensity of the sensor data based on the multidimensional difference feature vector; constructing a sampling parameter correlation function, which includes the mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal; the sampling parameters include sampling frequency and sampling accuracy; configuring actual sampling parameters for the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter correlation function, so that the sensor terminal can acquire data with the actual sampling parameters to obtain actual sensor data; updating the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data according to a preset time window, wherein if the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased, and if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased.

[0115] This application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0116] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0117] It should be noted that the technical solutions or features described in the above embodiments can be combined or supplemented with each other without conflict. The scope of protection of this application is not limited to the precise structures described in the above embodiments and shown in the accompanying drawings; all modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for acquiring and processing sensor data, characterized in that, include: The sensor data of each sensor terminal is analyzed and processed to obtain the time-domain difference features, frequency-domain difference features and spatial-domain difference features of the sensor data corresponding to each sensor terminal. Based on the time-domain difference features, the frequency-domain difference features and the spatial-domain difference features, a multi-dimensional difference feature vector of the sensor data corresponding to each sensor terminal is obtained. The intensity of change in the sensor data is calculated based on the multidimensional difference feature vector. Construct a sampling parameter association function, which includes the mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal, wherein the sampling parameters include sampling frequency and sampling accuracy; Based on the actual resource information of the sensor terminal and the sampling parameter association function, the actual sampling parameters are configured to the sensor terminal so that the sensor terminal can collect data using the actual sampling parameters to obtain actual sensor data. According to a preset time window, the actual sampling parameters of the sensor terminal are updated based on the actual change intensity of the actual sensor data. If the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased. Obtain the task metadata of the task corresponding to the actual sensor data; Identify the data set that the task depends on, and obtain the importance weight of each data in the data set to the task, thus obtaining an importance weight set; Based on the task metadata, the set of importance weights, and the preset data value model, the actual data value of the task is obtained; Whether to perform error correction processing on the actual sensor data is determined based on the actual data value. The resource information of the sensor terminal includes storage resource utilization, network bandwidth utilization, and computing resource utilization; the construction of the sampling parameter correlation function includes: Construct the sampling parameter correlation function based on the following formula: f_sample=f_base×(1+α·I_change)×(S_used / S_total)^(-β_s)×(B_used / B_total)^(-β_b)×(C_used / C_total)^(-β_c); p_sample=p_base×(1+γ·I_change)×(S_used / S_total)^(-δ_s)×(B_used / B_total)^(-δ_b); Where f_sample is the sampling frequency, p_sample is the sampling precision, f_base is the baseline sampling frequency, p_base is the baseline sampling precision, S_used / S_total is the storage resource utilization, I_change is the change intensity, B_used / B_total is the network bandwidth utilization, C_used / C_total is the computing resource utilization, α is the influence coefficient of change intensity on sampling frequency, γ is the influence coefficient of change intensity on sampling precision, β_s is the constraint coefficient of storage resource utilization on sampling frequency, β_b is the constraint coefficient of network bandwidth utilization on sampling frequency, β_c is the constraint coefficient of computing resource utilization on sampling frequency, δ_s is the constraint coefficient of storage resource utilization on sampling precision, and δ_b is the resource constraint coefficient of network bandwidth utilization on sampling precision.

2. The sensor data acquisition and processing method according to claim 1, characterized in that, The analysis and processing of sensor data from each sensor terminal yields temporal, frequency, and spatial difference characteristics of the corresponding sensor data for each sensor terminal, including: Calculate the difference between adjacent sampling points of the sensor data in the time domain to obtain the time-domain difference characteristics; Perform a fast Fourier transform on the sensor data to obtain the spectrum, extract spectral features from the spectrum and process them to obtain frequency domain difference features; Spatial correlation parameters of sensor data from adjacent sensor terminals in the spatial dimension are calculated to obtain spatial domain difference characteristics.

3. The sensor data acquisition and processing method according to claim 1, characterized in that, The step of calculating the change intensity of the sensor data based on the multidimensional difference feature vector includes: The multidimensional difference feature vector is processed using a normalization function to obtain the intensity of change of the value in the interval between 0 and 1.

4. The sensor data acquisition and processing method according to claim 1, characterized in that, The function for constructing the sampling parameter correlation also includes: A reinforcement learning model is used to optimize the data change influence coefficient and resource constraint coefficient in the correlation function of the sampling parameters.

5. The sensor data acquisition and processing method according to claim 1, characterized in that, Also includes: Establish a data acquisition quality evaluation model to obtain the sensor data for the time period to be evaluated; The data integrity score, accuracy score, and timeliness score of the sensor data for the period to be evaluated are obtained through the data acquisition quality evaluation model. Based on the data integrity score, the accuracy score, and the timeliness score, a comprehensive acquisition quality score is obtained for the sensor data corresponding to the period to be evaluated.

6. The sensor data acquisition and processing method according to claim 1, characterized in that, Also includes: After determining that the actual sensor data should be corrected, abnormal data detection is performed on the actual sensor data to remove abnormal data points and obtain cleaned sensor data. Based on the cleaned sensor data, interpolation is performed on the abnormal and missing data points in the actual sensor data to obtain the corrected sensor data.

7. The sensor data acquisition and processing method according to claim 6, characterized in that, The step of detecting abnormal data in the actual sensor data and removing abnormal data points to obtain cleaned sensor data includes: The actual sensor data is encoded and compressed using a bidirectional long short-term memory network encoder to obtain compressed data. The compressed data is reconstructed using a bidirectional long short-term memory network decoder to obtain the reconstructed data; Calculate the reconstruction error between the reconstructed data and the actual sensor data; Data points whose reconstruction error exceeds a preset anomaly threshold are marked as anomaly data points; The abnormal data points are filtered out from the actual sensor data to obtain cleaned sensor data.

8. The sensor data acquisition and processing method according to claim 7, characterized in that, The step of interpolating abnormal and missing data points in the actual sensor data based on the cleaned sensor data to obtain corrected sensor data includes: A multi-layer convolutional neural network model is used to extract spatial correlation features of the sensor terminal; A multi-layer long short-term memory network model was used to extract the temporal correlation features of the cleaned sensor data. A nonlinear interpolation function is constructed based on the cleaned sensor data, the spatial correlation features, and the temporal correlation features; Based on the nonlinear interpolation function, abnormal data points and missing data points in the actual sensor data are interpolated to obtain the error-corrected sensor data.

9. A sensor data acquisition and processing device, characterized in that, include: The multidimensional difference feature vector acquisition module is used to analyze and process the sensor data of each sensor terminal to obtain the time domain difference features, frequency domain difference features and spatial domain difference features of the sensor data corresponding to each sensor terminal, and to obtain the multidimensional difference feature vector of the sensor data corresponding to each sensor terminal based on the time domain difference features, the frequency domain difference features and the spatial domain difference features. The change intensity acquisition module is used to calculate the change intensity of the sensor data based on the multidimensional difference feature vector. The correlation function construction module is used to construct a sampling parameter correlation function, which includes the mapping relationship between the sampling parameters of the sensor terminal and the change intensity and the resource information of the sensor terminal. The sampling parameters include sampling frequency and sampling accuracy. The actual sampling parameter configuration module is used to configure actual sampling parameters to the sensor terminal based on the actual resource information of the sensor terminal and the sampling parameter association function, so that the sensor terminal can collect data using the actual sampling parameters to obtain actual sensor data. The dynamic adjustment module is used to update the actual sampling parameters of the sensor terminal according to the actual change intensity of the actual sensor data within a preset time window. If the actual change intensity is greater than a first threshold, the value of the actual sampling parameter is increased; if the actual change intensity is less than a second threshold, the value of the actual sampling parameter is decreased. The task element extraction module is used to obtain the task element information of the task corresponding to the actual sensor data; The data dependency analysis module is used to identify the data set that the task depends on, and to obtain the importance weight of each data in the data set to the task, thereby obtaining an importance weight set. The data value modeling module is used to obtain the actual data value of the task based on the task metadata, the set of importance weights, and a preset data value model. The error correction decision module is used to determine whether to perform error correction processing on the actual sensor data based on the actual data value. The resource information of the sensor terminal includes storage resource utilization, network bandwidth utilization, and computing resource utilization; the correlation function construction module is configured as follows: Construct the sampling parameter correlation function based on the following formula: f_sample=f_base×(1+α·I_change)×(S_used / S_total)^(-β_s)×(B_used / B_total)^(-β_b)×(C_used / C_total)^(-β_c); p_sample=p_base×(1+γ·I_change)×(S_used / S_total)^(-δ_s)×(B_used / B_total)^(-δ_b); Where f_sample is the sampling frequency, p_sample is the sampling precision, f_base is the baseline sampling frequency, p_base is the baseline sampling precision, S_used / S_total is the storage resource utilization, I_change is the change intensity, B_used / B_total is the network bandwidth utilization, C_used / C_total is the computing resource utilization, α is the influence coefficient of change intensity on sampling frequency, γ is the influence coefficient of change intensity on sampling precision, β_s is the constraint coefficient of storage resource utilization on sampling frequency, β_b is the constraint coefficient of network bandwidth utilization on sampling frequency, β_c is the constraint coefficient of computing resource utilization on sampling frequency, δ_s is the constraint coefficient of storage resource utilization on sampling precision, and δ_b is the resource constraint coefficient of network bandwidth utilization on sampling precision.

10. The sensor data acquisition and processing device according to claim 9, characterized in that, Also includes: An anomaly detection module is used to detect anomalies in the actual sensor data after determining that error correction processing should be performed on the actual sensor data, remove abnormal data points in the actual sensor data, and obtain cleaned sensor data. The error correction module is used to interpolate abnormal and missing data points in the actual sensor data based on the cleaned sensor data to obtain error-corrected sensor data.

11. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory being communicatively connected to the processor, and the memory storing instructions. The processor invokes the instructions in the memory, causing the electronic device to execute the sensor data acquisition and processing method according to any one of claims 1 to 8.

12. A computer-readable storage medium storing instructions thereon, characterized in that, When the instruction is executed by the processor, it implements the sensor data acquisition and processing method according to any one of claims 1 to 8.

Images