Multi-source heterogeneous environment data credibility evaluation method and system based on kernel density estimation

Abstract

The present application relates to the technical field of new energy power generation environment monitoring, and particularly relates to a multi-source heterogeneous environment data credibility evaluation method and system based on kernel density estimation, wherein temperature, humidity and wind speed environment data are collected by a plurality of deployed sensors; the collected data is subjected to outlier detection, outlier correction and data filtering processing; a known accurate sample is modeled by using a kernel density estimation method to obtain an accurate data probability density distribution, and the density value of the collected new data under the distribution is calculated as an accuracy index; the variance of each type of environment data sequence is calculated as a stability index; the data consistency is evaluated by using KL divergence; and the credibility of each specific data value of each type of data is calculated according to the weight distribution of the accuracy, stability and consistency indexes. The present application realizes high-precision credibility evaluation of environment sensor data by using kernel density estimation and multi-source data fusion technology, thereby improving the monitoring reliability and decision accuracy of the new energy power generation system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy power generation environmental monitoring, particularly relates to a multi-source heterogeneous environmental data credibility evaluation method and system based on kernel density estimation, which evaluates the credibility of temperature, humidity, wind speed and other environmental factor data collected by multiple sensors. BACKGROUND

[0002] In the field of new energy power generation, accurate monitoring and credibility evaluation of environmental factor data are crucial for power generation prediction and optimization. The power generation efficiency of new energy sources such as solar and wind energy is directly affected by environmental conditions, for example, changes in environmental factors such as light intensity, temperature, and wind speed can significantly affect the output power of solar panels and wind turbines.

[0003] Therefore, it is necessary to collect these environmental factor data in real time through multiple sensors and accurately evaluate their credibility, which can help improve the accuracy of power generation prediction and optimize the operation and management of new energy power generation systems. Traditional data evaluation methods have limitations in handling such multi-source heterogeneous data, which may not fully consider the complex fluctuations and correlations of data, resulting in inaccurate evaluation results. Kernel density estimation, as a flexible non-parametric probability density estimation method, can better adapt to the actual distribution of data and provide more reliable support for the credibility evaluation of environmental factor data.

[0004] Traditional data evaluation methods have the following shortcomings when dealing with multi-source heterogeneous data:

[0005] Inadequate consideration of complex fluctuations in data: Traditional data evaluation methods may not be able to adapt to the complex fluctuation characteristics of environmental data, resulting in inaccurate evaluation results.

[0006] Inadequate consideration of data correlation: Traditional data evaluation methods may ignore the correlation between different environmental factor data, affecting the comprehensiveness and accuracy of the evaluation.

[0007] Inaccurate evaluation results: Due to the above two reasons, traditional data evaluation methods have deficiencies in the accuracy and reliability of overall evaluation results when dealing with multi-source heterogeneous environmental data. SUMMARY

[0008] The present application aims to solve the problem of traditional data evaluation methods not fully considering the complex fluctuations and correlations of data when dealing with multi-source heterogeneous environmental data, resulting in inaccurate evaluation results. A multi-source heterogeneous environmental data credibility evaluation method and system based on kernel density estimation is proposed, which uses the flexible non-parametric probability density estimation method of kernel density estimation to better adapt to the actual distribution of data, thereby providing more reliable support for the credibility evaluation of environmental factor data.

[0009] To achieve the above object, the technical scheme adopted is:

[0010] The application provides a multi-source heterogeneous environment data credibility evaluation method based on kernel density estimation, comprising the following steps:

[0011] Step 1, collecting temperature T, humidity H and wind speed W environment data through a plurality of deployed sensors, ensuring time synchronization and format uniformity of the data;

[0012] Step 2, performing abnormal value detection, abnormal value correction and data filtering processing on the collected data;

[0013] Step 3, modeling the known accurate samples by using the kernel density estimation method, obtaining the accurate data probability density distribution, and calculating the density value of the collected new data under the distribution as the accuracy index;

[0014] Step 4, calculating the variance of each type of environment data sequence as the stability index;

[0015] Step 5, evaluating the data consistency by using the Kullback-Leibler divergence;

[0016] Step 6, comprehensively evaluating the accuracy, stability and consistency indexes, and calculating the credibility of each specific data value of each type of data according to the weight distribution.

[0017] According to the multi-source heterogeneous environment data credibility evaluation method based on kernel density estimation, further, the time synchronization of the data in step 1 is ensured by using a precise time protocol clock synchronization server, and the time synchronization between the sensors is realized by receiving an external high-precision clock source signal, thereby ensuring the consistency of the data timestamp.

[0018] According to the multi-source heterogeneous environment data credibility evaluation method based on kernel density estimation, further, the abnormal value detection of the collected data in step 2 is performed by using a statistical method, specifically: calculating the mean and standard deviation of the data sequence, and regarding the data beyond the range of ±3 times the standard deviation of the mean as abnormal values.

[0019] According to the multi-source heterogeneous environment data credibility evaluation method based on kernel density estimation, further, the abnormal value correction of the collected data in step 2 is performed by using a neighboring data point mean method, specifically: for the detected abnormal value, the mean of the four data points is calculated, and the mean is used to replace the abnormal value; if the first data is abnormal, the previous two data are considered as 0; if the last data is abnormal, the last two data are considered as 0.

[0020] According to the reliability assessment method for multi-source heterogeneous environmental data based on kernel density estimation of the present invention, in step 2, the data filtering adopts the median filtering method, specifically: the median of the data within the sliding window is taken as the value of the noise point to remove impulse noise.

[0021] According to the reliability assessment method for multi-source heterogeneous environmental data based on kernel density estimation of the present invention, step 3 further includes:

[0022] Using kernel density estimation to model known accurate historical data, we obtain the probability density function. X is any sequence of sensor data;

[0023] To evaluate the current data point X i Calculate its probability density in the distribution.

[0024] The calculated density value Divide by the maximum density value of all data points in the current sequence The normalized accuracy is obtained; the closer the value is to 1, the more the data point conforms to the normal distribution.

[0025] According to the multi-source heterogeneous environmental data reliability assessment method based on kernel density estimation of the present invention, step 5, which uses Kullback-Leibler divergence to assess data consistency, specifically involves:

[0026] Kernel density estimation was performed on the temperature, humidity, and wind speed data to obtain univariate distributions.

[0027] Calculate the KL divergence between pairwise distributions

[0028] The average KL divergence of the same sensor is taken to obtain the average KL value for data consistency assessment of that sensor. X The smaller this value, the more consistent the sensor data is with other sensor data.

[0029] According to the reliability assessment method for multi-source heterogeneous environmental data based on kernel density estimation of the present invention, the reliability calculation formula in step 6 is further as follows:

[0030]

[0031] Where w1, w2, w3 are weight coefficients, satisfying w1 + w2 + w3 = 1; Indicates the current data point X i The stability of the data sequence to which it belongs; all data points in the same sequence share the same stability.

[0032] According to the multi-source heterogeneous environment data credibility evaluation method based on kernel density estimation, further, the weight coefficient is determined by expert scoring method, the accuracy weight w1=0.4, the stability weight w2=0.3, and the consistency weight w3=0.3.

[0033] Further, the present application also provides a multi-source heterogeneous environment data credibility evaluation system based on kernel density estimation, for realizing the multi-source heterogeneous environment data credibility evaluation method based on kernel density estimation, the system comprises:

[0034] A data acquisition module is configured to acquire temperature T, humidity H, and wind speed W environment data through a plurality of deployed sensors, and ensure time synchronization and format uniformity of the data.

[0035] A preprocessing module is configured to perform outlier detection, outlier correction, and data filtering processing on the acquired data.

[0036] An accuracy estimation module is configured to model known accurate samples using a kernel density estimation method, obtain an accurate data probability density distribution, and calculate a density value of the collected new data under the distribution as an accuracy index.

[0037] A stability evaluation module is configured to calculate the variance of each type of environment data sequence as a stability index.

[0038] A consistency evaluation module is configured to evaluate data consistency using Kullback-Leibler divergence.

[0039] A credibility calculation module is configured to comprehensively consider the accuracy, stability, and consistency indexes, and calculate the credibility of each specific data value of each type of data according to the weight distribution.

[0040] The above technical solution has the following beneficial effects:

[0041] 1. Improve the accuracy and reliability of data credibility evaluation

[0042] The present application uses non-parametric kernel density estimation (KDE) to model data distribution, which can better adapt to the actual distribution of environment data, effectively capture the real characteristics of complex fluctuations and multi-source heterogeneous data, overcome the limitations of traditional parameter methods, and thus more accurately evaluate the credibility of multi-source heterogeneous environment data.

[0043] 2. Universal and scalable

[0044] The present application is suitable for various environment monitoring scenarios, especially for the new energy power generation field. The method and model have universality and can be extended to other similar multi-source data evaluation scenarios (such as transportation), and have strong adaptability.

[0045] 3. Optimizing operation and management of new energy power generation system

[0046] By accurately evaluating the reliability of environmental factor data, the accuracy of power generation prediction can be improved, thereby optimizing the operation and management of the new energy power generation system, which has practical application value.

[0047] 4. Providing quantitative evaluation indexes

[0048] The present application not only evaluates the reliability of data, but also provides quantifiable evaluation indexes, making the evaluation results more intuitive, objective and easy to apply.

[0049] 5. Comprehensive consideration of multiple factors

[0050] The reliability value is calculated by weight distribution from multiple dimensions such as data accuracy, stability and consistency, ensuring the comprehensiveness and scientificity of the evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application will be briefly introduced below. Among them, the drawings are only used to show some embodiments of the present application, and not to limit all embodiments of the present application to this.

[0052] Figure 1 is a flowchart of the multi-source heterogeneous environmental data reliability evaluation method based on kernel density estimation of the embodiments of the present application;

[0053] Figure 2 is a data transmission process schematic diagram of the embodiments of the present application. DETAILED DESCRIPTION

[0054] The example solutions of the embodiments of the present application will be described clearly and completely below in combination with the drawings of the specific embodiments of the present application. Unless otherwise defined, the technical terms or scientific terms used in the present application should be understood as the usual meaning understood by persons having ordinary skills in the art.

[0055] The present embodiment discloses a multi-source heterogeneous environmental data reliability evaluation method based on kernel density estimation, which improves the evaluation accuracy and reliability, as shown in Figure 1 The method includes data acquisition, preprocessing, accuracy evaluation, stability evaluation, consistency evaluation, reliability calculation and result output steps. Specifically, it includes steps S1-S7:

[0056] Step S1, collect temperature T, humidity H, wind speed W environmental data through the deployed multiple sensors, ensure the time synchronization and format uniformity of the data.

[0057] As shown in Figure 2As shown, the time synchronization server is used for time synchronization between various sensors to maintain the consistency of time scale values ​​of various sensors. Sensor data is transmitted to the host computer via a switch. The host computer software runs on a general-purpose computer and is responsible for outlier detection, correction, filtering, and reliability assessment of the data. Step S1 includes sub-steps S101-S104.

[0058] Step S101, Sensor Deployment

[0059] Sensor deployment: Temperature, humidity, and wind speed sensors should be deployed appropriately in the environmental monitoring area of ​​the new energy photovoltaic power plant to ensure that the data is representative.

[0060] For example, temperature sensors should be installed about 0.5 meters below the photovoltaic panels; humidity sensors should be placed in the ventilation channels between the photovoltaic panels, about 1.5 meters above the ground, and should be protected from direct sunlight; wind speed sensors should be installed in an open, unobstructed area within the photovoltaic power station, at a height of about 10 meters above the ground.

[0061] Step S102, Sensor clock synchronization

[0062] Data synchronization: Time synchronization technology is used to ensure the consistency of data acquisition time from multiple sensors, thereby ensuring the synchronization of timestamps within the reported numerical messages.

[0063] The PTP (Precise Time Protocol) clock synchronization server serves as the reference clock source for the entire system. It receives high-precision time signals from external clock sources (such as GPS and BeiDou) and generates a precise reference clock. Following the IEEE 1588 series of protocols, the server periodically sends time synchronization messages to various sensors in the network. These messages contain precise timestamp information, which the sensors use to adjust their own clocks.

[0064] After receiving a time synchronization message from the PTP clock synchronization server, the sensor calculates the time deviation based on the time difference between the timestamp in the message and its own clock, and uses this deviation to adjust its own clock to achieve synchronization with the reference clock. To ensure the accuracy and reliability of synchronization, the sensor needs to implement a precise timestamp marking function to accurately record the timestamp during data acquisition.

[0065] Step S103, Data Acquisition Format

[0066] Data format: The collected data should include information such as sensor identification, environmental factor type, data value, and timestamp.

[0067] The following is an example of data acquisition, using temperature as an example:

[0068]

[0069]

[0070] Step S104, data collection content

[0071] Collect temperature, humidity, and wind speed data through three different sensors, and collect q data for each type of data. Each sensor collects data once per millisecond, and the temperature data sequence obtained within q milliseconds is T = {T1, T2,... Tq}, the humidity data sequence is H = {H1, H2,... Hq}, and the wind speed data sequence is W = {W1, W2,... Wq}. q}。 q q

[0072] Step S2, preliminary processing of collected data, including outlier detection, outlier correction, and data filtering.

[0073] Step S201, outlier detection

[0074] Detect outliers using statistical methods, calculate the mean and standard deviation, and consider data outside the range of mean ± 3 times the standard deviation as outliers.

[0075] Taking temperature as an example:

[0076] For the temperature data sequence T, calculate its mean μ T and standard deviation σ T , and determine whether each data point T i is within the range [μ T - 3μ T , μ T + 3σ T ].

[0077] The formula for calculating the mean is:

[0078]

[0079] The formula for calculating the standard deviation is:

[0080]

[0081] Step S202, outlier correction

[0082] Taking the temperature data sequence T as an example:

[0083] Correct the outliers using the mean of adjacent data points. For the detected outliers T i ​​, take the first two and the last two data points (a total of four data points), calculate the mean of the four data points, and replace the abnormal value T with the mean i The correction formula is:

[0084]

[0085] If the first data is abnormal, the first two data are considered to be 0; if the last data is abnormal, the last two data are considered to be 0.

[0086] Since the first data point has no predecessor data (i.e. no data point in front), it is impossible to obtain the "first two data points", so the "first two data points" are considered to be 0, and only the mean of the last two data points is used to correct the abnormal value, so the correction formula degenerates to:

[0087]

[0088] Since the last data point has no successor data (i.e. no data point behind), it is impossible to obtain the "last two data points", so the "last two data points" are considered to be 0, and only the mean of the first two data points is used to correct the abnormal value, and the correction formula degenerates to:

[0089]

[0090] It should be noted that a certain data that is replaced due to abnormality has a confidence of 0 at the same time, and other source data also has a confidence of 0. Therefore, even if the data of other sensors (such as humidity and wind speed) appears to be normal, the confidence is forced to be zero, avoiding potential false data from polluting subsequent analysis.

[0091] Step S203, numerical filtering

[0092] The median filtering method is adopted to take the median of the data in a period of time to remove impulse noise. The formula is:

[0093] T i ' = median (T i-N / 2 , T i-N / 2+1 ,..., T i+N / 2 )

[0094] Where T i ' is the filtered data, N is the filter window size, usually an odd number, and N = 5 is taken in this embodiment.

[0095] It should be noted that a certain data that is filtered out has a confidence of 0 at the same time, and other source data also has a confidence of 0.

[0096] Step S3, modeling the known accurate samples using kernel density estimation method to obtain the accurate data probability density distribution, calculating the density value of the collected new data under the distribution as the accuracy index. Step S3 includes sub-steps S301-S305.

[0097] Step S301, data modeling

[0098] Modeling the known accurate historical data using kernel density estimation to obtain the probability density function X is an arbitrary sensor data sequence.

[0099] Taking the temperature sensor data sequence T as an example, h is the bandwidth, which controls the width of the kernel function and thus affects the smoothness of the estimation.

[0100] Probability density of bandwidth h As follows:

[0101]

[0102] Where K(·) is the kernel function, and the Gaussian kernel function is selected:

[0103]

[0104] If T1=20 in the temperature sensor data sequence T, the bandwidth h=3, and the total data amount of the sequence is 100, then the probability density of T1=20 is:

[0105]

[0106] Step S302, bandwidth h determination method

[0107] The bandwidth h has a large impact on the result, and the preliminary estimation formula is:

[0108]

[0109] Where σ is the standard deviation of the data, and q is the sample size, and then optimized through cross-validation.

[0110] Cross-validation divides the data set into a training set and a validation set (taking the temperature sensor data sequence T as an example, the first 50% data is selected as the training set, and the last 50% data is selected as the validation set), and a series of candidate bandwidth values h1, h2,..., h m are randomly selected according to experience. For each candidate bandwidth h s , the training set data is used for kernel density estimation to obtain the density estimation function

[0111] The mean square error on the validation set is calculated, and the bandwidth that minimizes the mean square error is selected. The formula is:

[0112]

[0113] where q is the number of samples, 0.5q is the validation set size, is the estimated density with bandwidth h s , f(T i ) is the true density.

[0114] Step S303, calculate single point density value

[0115] For the current data point T i to be evaluated, calculate its probability density in the distribution The higher the density value, the more likely T i belongs to the normal data distribution.

[0116] Step S304, normalize to accuracy

[0117] Divide the calculated density value by the maximum density value of all data points in the current sequence to get the normalized accuracy:

[0118]

[0119] The closer this value is to 1, the higher the accuracy of the data, indicating that the data is more consistent with the normal distribution.

[0120] Step S305, update the historical data sequence with a sliding window and re-fit to adapt to environmental changes.

[0121] Step S4, calculate the variance of each type of environmental data sequence as a stability index.

[0122] Calculate the variance of each type of data sequence as a stability index, and the smaller the value, the more stable the data.

[0123] Take the temperature sensor data sequence T as an example:

[0124] Variance calculation formula:

[0125]

[0126] where μ T is the data mean:

[0127]

[0128] It should be noted that when calculating the credibility of a data point T i , the variance of the entire temperature sensor data sequence T is directly used as its stability input. For example, if the sequence variance σ T 2 = 4, then all Ti In the credibility calculation, the stability term is always

[0129] Step S5: Use Kullback-Leibler divergence to assess data consistency, or estimate the joint distribution of kernel density after mixing data from different sensors, and calculate the density value of each sensor data under this distribution. The closer the values ​​are, the higher the data consistency.

[0130] This example uses Kullback-Leibler divergence to assess consistency.

[0131] First, calculate the KL divergence between the probability distributions of every two data sequences, and then calculate the average KL dispersion value.

[0132] The probability density function of temperature sensing data sequence T The probability density function of the wind speed sensor data sequence W and These are the kernel density estimation functions for data from different sensors. The smaller the KL divergence, the more similar the distributions, and the higher the data consistency.

[0133] and The KL dispersion between them is calculated as follows:

[0134]

[0135] Calculated and The KL dispersion between the temperature sensor data sequence T and the humidity sensor data sequence H is calculated, and then the KL dispersion between their probability density functions is calculated. The weighted average is then used to obtain the consistency assessment of the temperature sensor data.

[0136] Let KL be the humidity probability density function. T This is the average value for consistency assessment of temperature sensing data. The closer the value is to 1, the higher the consistency of the data.

[0137]

[0138] Understandably, consistency metrics for each sensor type, such as temperature consistency (KL), are important. T Used for all data points T in the temperature sensing data sequence T i Calculation of credibility.

[0139] Step S6: Based on the comprehensive accuracy, stability, and consistency indices, calculate the reliability of each specific data value for each data category according to the weighted allocation.

[0140] The accuracy, stability and consistency indexes of the three types of data of temperature, humidity and wind speed are allocated weights, and the credibility value C of each specific data value of each type of data is calculated, such as the credibility value C of the temperature T i The value formula is:

[0141]

[0142] Where w1, w2, w3 are weight coefficients, and w1 + w2 + w3 = 1, such as w1 = 0.4, w2 = 0.3, w3 = 0.3, and the weights can be adjusted according to the actual scene and requirements, and the expert scoring method can be used to determine the weight of each index. The stability of the data sequence to which the current data point T i belongs, and the value is closer to 1, indicating that the data is more stable. The use of ensures that the stability is positively correlated with the credibility (the smaller the fluctuation, the higher the value, and the higher the credibility). The denominator max(σ T 2 ) of the stability term is based on historical data. The denominator max(σ ) of the accuracy term is based on the maximum density value of the current analyzed data sequence (new data).

[0143] Step S7, result output and application

[0144] The credibility value of each specific data value of each type of data is output, and the demand is filtered, fused or supported for decision-making, such as selecting data with high credibility for environmental analysis and decision-making to optimize new energy power generation prediction and system operation.

[0145] Corresponding to the above method, the embodiment also discloses a multi-source heterogeneous environmental data credibility evaluation system based on kernel density estimation, comprising:

[0146] A data acquisition module is configured to acquire temperature T, humidity H and wind speed W environmental data through a plurality of deployed sensors, and ensure time synchronization and format uniformity of the data.

[0147] A preprocessing module is configured to detect, correct and filter the collected data.

[0148] An accuracy estimation module is configured to model known accurate samples using a kernel density estimation method, obtain an accurate data probability density distribution, calculate the density value of the collected new data under the distribution, and use the density value as an accuracy index.

[0149] A stability evaluation module is configured to calculate the variance of each type of environmental data sequence as a stability index.

[0150] A consistency evaluation module is configured to evaluate data consistency by using Kullback-Leibler divergence.

[0151] A credibility calculation module is configured to calculate credibility of each specific data value of each type of data according to weight distribution by integrating accuracy, stability and consistency indexes.

[0152] Unless otherwise specified, the relative steps, numerical expressions and values of components and steps set forth in these embodiments do not limit the scope of the present application.

[0153] The various embodiments are described in the present specification in a progressive manner, each of which focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the system disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0154] The units and method steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been described in the above description in general terms. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation does not exceed the scope of the present application.

[0155] Those skilled in the art can understand that all or part of the steps in the above method can be instructed by a program to complete the relevant hardware, and the program can be stored in a computer readable storage medium, such as a read-only memory, a magnetic disk or an optical disk. Alternatively, all or part of the steps of the above embodiments can also be implemented using one or more integrated circuits, and accordingly, each module / unit in the above embodiments can be implemented in the form of hardware or in the form of a software function module. The present application is not limited to any specific form of combination of hardware and software.

[0156] Finally, it should be noted that the above-described embodiments are merely specific embodiments of the present application, which are used to illustrate the technical solutions of the present application, but not to limit the same. The protection scope of the present application is not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that any person skilled in the art can still modify or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some of the technical features, within the technical scope disclosed by the present application. The modifications, changes or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation, characterized in that, Includes the following steps: Step 1: Collect environmental data such as temperature (T), humidity (H), and wind speed (W) using multiple deployed sensors to ensure data synchronization in time and consistency in format; Step 2: Perform outlier detection, outlier correction, and data filtering on the collected data; Step 3: Model the known accurate samples using the kernel density estimation method to obtain the probability density distribution of the accurate data. Calculate the density value of the newly collected data under this distribution as an accuracy indicator. This specifically includes: Using kernel density estimation to model known accurate historical data, we obtain the probability density function. X is any sequence of sensor data; To evaluate the current data point X i Calculate its probability density in the distribution. The calculated density value Divide by the maximum density value of all data points in the current sequence The normalized accuracy is obtained; the closer the value is to 1, the more the data point conforms to the normal distribution. Step 4: Calculate the variance of each type of environmental data sequence as a stability index; Step 5: Use Kullback-Leibler divergence to assess data consistency, specifically including: Kernel density estimation was performed on the temperature, humidity, and wind speed data to obtain univariate distributions. Calculate the KL divergence between pairwise distributions The average KL divergence of the same sensor is taken to obtain the average KL value for data consistency assessment of that sensor. X The smaller this value, the more consistent the sensor data is with data from other sensors. Step 6: Based on the comprehensive accuracy, stability, and consistency indicators, calculate the reliability of each specific data value for each data category according to the weighting.

2. The method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation according to claim 1, characterized in that, In step 1, the time synchronization of the data is ensured by using a precise time protocol clock synchronization server. This server receives external high-precision clock source signals to achieve time synchronization between sensors and ensure the consistency of data timestamps.

3. The method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation according to claim 1, characterized in that, In step 2, outlier detection of the collected data is performed using statistical methods, specifically: the mean and standard deviation of the data sequence are calculated, and data exceeding the mean ± 3 times the standard deviation are considered outliers.

4. The method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation according to claim 1, characterized in that, In step 2, the outlier correction of the collected data is performed using the mean of nearest neighbor data points. Specifically, for each detected outlier, the two preceding and two following data points are taken, the mean of these four data points is calculated, and the outlier is replaced with the mean. If the first data point is outlier, the first two data points are considered to be 0. If the last data point is outlier, the last two data points are considered to be 0.

5. The method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation according to claim 1, characterized in that, In step 2, the data filtering uses median filtering, specifically: the median of the data within the sliding window is taken as the value of the noise point to remove impulse noise.

6. The method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation according to claim 1, characterized in that, The formula for calculating credibility in step 6 is: Where w1, w2, w3 are weight coefficients, satisfying w1 + w2 + w3 = 1; Indicates the current data point X i The stability of the data sequence to which it belongs; all data points in the same sequence share the same stability.

7. The method for assessing the credibility of multi-source heterogeneous environmental data based on kernel density estimation according to claim 6, characterized in that, The weighting coefficients are determined by expert scoring, with accuracy weight w1 = 0.4, stability weight w2 = 0.3, and consistency weight w3 = 0.

3.

8. A reliability assessment system for multi-source heterogeneous environmental data based on kernel density estimation, characterized in that, For implementing the multi-source heterogeneous environmental data credibility assessment method based on kernel density estimation as described in any one of claims 1-7, the system comprises: The data acquisition module is used to collect environmental data such as temperature (T), humidity (H), and wind speed (W) through multiple deployed sensors, ensuring the time synchronization and format consistency of the data. The preprocessing module is used to perform outlier detection, outlier correction, and data filtering on the collected data; The accuracy estimation module is used to model known accurate samples using the kernel density estimation method, obtain the probability density distribution of accurate data, and calculate the density value of newly collected data under this distribution as an accuracy index. The stability assessment module is used to calculate the variance of each type of environmental data sequence as a stability index. The consistency assessment module is used to assess data consistency using Kullback-Leibler divergence. The credibility calculation module is used to comprehensively evaluate accuracy, stability, and consistency indicators, and calculate the credibility of each specific data value for each data category according to weighted allocation.

Images