Internet of Things sensor data fusion method for smart city

Abstract

The invention relates to the field of Internet of Things data processing, in particular to an Internet of Things sensor data fusion method for a smart city. The method comprises the steps of collecting sensor data and performing standardization processing to obtain a standardized data set; preprocessing the standardized data set through a low-pass filtering-dynamic smoothing technology to obtain a smooth data set; dynamically adjusting the size of the sliding window through an improved dynamic adaptive window adjustment algorithm, extracting a sliding window data set from the smooth data set based on the size of the sliding window, and performing data fusion on the sliding window data set to obtain a fusion estimation result of the sliding window data; and generating a comprehensive estimation result of the sliding window data in combination with the fusion estimation result of all the data in the sliding window. The problem of key information loss possibly caused by high-frequency filtering and average smoothing methods is solved; the size of a fixed sliding window may cause inaccurate data fusion; and when diversified and nonlinear data is handled, the calculation overhead is relatively high, and the real-time performance is insufficient.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things data processing, and in particular to an Internet of Things sensor data fusion method for a smart city. Background Art

[0002] With the construction and development of smart cities, the widespread application of IoT sensor technology has provided a large amount of data support for the intelligent management of cities. Traditional data processing methods usually rely on static filtering and fixed window technology, which are mainly used to remove noise and extract data features. However, with the increase in data complexity and dynamics, simple filtering and static analysis methods are no longer sufficient to cope with diverse and nonlinear data. Therefore, dynamic adaptive methods, including dynamic smoothing and adaptive window adjustment, are gradually introduced to better capture the changing trends of data and improve the accuracy and real-time performance of data processing.

[0003] However, the above data processing technology has the following technical problems: there are certain limitations when processing high-noise and complex-changing data. High-frequency filtering and average smoothing methods may cause key information loss while removing noise; the fixed sliding window size may not be able to adapt to the dynamic changes of data, resulting in a decrease in the accuracy of data fusion results; when dealing with diverse and nonlinear data, it faces the problems of high computational overhead and lack of real-time performance; these problems seriously restrict the efficiency and accuracy of IoT sensor data fusion in smart cities, and a more flexible and adaptive fusion method is urgently needed. Summary of the invention

[0004] The present invention provides an Internet of Things sensor data fusion method for smart cities, so as to solve the problems that high-frequency filtering and average smoothing methods may cause key information loss while removing noise; the fixed sliding window size may not be able to adapt to the dynamic changes of data, resulting in a decrease in the accuracy of data fusion results; and when dealing with diversified and nonlinear data, the calculation overhead is large and the real-time performance is insufficient.

[0005] The present invention provides a method for fusion of IoT sensor data for smart cities, which specifically includes the following technical solutions:

[0006] A method for fusion of IoT sensor data for smart cities, comprising the following steps:

[0007] S1: Collect sensor data and perform standardization on the sensor data to obtain standardized data; obtain a standardized data set based on the standardized data; preprocess the standardized data set through low-pass filtering-dynamic smoothing technology to obtain a smoothed data set; dynamically adjust the sliding window size through an improved dynamic adaptive window adjustment algorithm, extract data from the smoothed data set based on the sliding window size, and obtain a sliding window data set;

[0008] S2: The sliding window data set is fused through a dynamic adaptive distributed fusion algorithm to obtain a fusion estimation result of the sliding window data; the fusion estimation results of all data in the sliding window are combined to generate a comprehensive estimation result of the sliding window data.

[0009] Preferably, the S1 specifically includes:

[0010] In the process of implementing the low-pass filtering-dynamic smoothing technology, a low-pass filter is applied, Gaussian weights and logarithmic functions are introduced, and the standardized data set is filtered to obtain a filtered data set.

[0011] Preferably, the S1 specifically includes:

[0012] Through the dynamic smoothing formula, the logarithmic and S-shaped functions are introduced to dynamically smooth the filtered data set to obtain a smoothed data set. The dynamic smoothing formula is as follows:

[0013]

[0014] Among them, d″ i is the i-th data in the smoothed data set; α is the smoothing coefficient; d′ i is the i-th data in the filtered data set; d″ i-1 is the i-1th data in the smoothed data set; is a sigmoid function.

[0015] Preferably, the S1 specifically includes:

[0016] In the implementation process of the improved dynamic adaptive window adjustment algorithm, based on the sliding window data set, the trend value of the sliding window data set is calculated, and the calculation formula is as follows:

[0017]

[0018] Where T is the trend value of the sliding window data set; W0 is the initial sliding window size; d″ v is the vth data in the sliding window data set; v is the index of the sliding window data; ∈ is a constant that prevents the denominator from being zero and the logarithmic calculation value from being zero.

[0019] Preferably, the S1 specifically includes:

[0020] Based on the trend value of the sliding window data set, the sliding window data change rate is calculated; based on the sliding window data change rate, the current sliding window size is adjusted, and the sliding window data set is updated; the sliding window data change rate is compared with a preset change rate threshold, and when the sliding window data change rate is less than or equal to the change rate threshold, or reaches a preset maximum number of adjustments, the sliding window data set is stopped from being updated; when the sliding window data set stops updating, the next sliding window step size is set based on the sliding window data change rate.

[0021] Preferably, the S2 specifically includes:

[0022] In the process of implementing the dynamic adaptive distribution fusion algorithm, the probability distribution function of the current sliding window data is calculated by the kernel density estimation formula; based on the probability distribution function of the current sliding window data, the distribution dynamics measure of the current sliding window data is calculated.

[0023] Preferably, the S2 specifically includes:

[0024] Based on the distribution dynamics measure of the current sliding window data, the weight of the current sliding window data is calculated.

[0025] Preferably, the S2 specifically includes:

[0026] Based on the weight of the current sliding window data and the probability distribution function of the current sliding window data, the current sliding window data set is adaptively weighted fused to obtain the fusion estimation result of the current sliding window data; the specific formula is as follows:

[0027] F(d″ v )=w v ·P(d″ v )·log(1+|w v ·P(d″ v )|)

[0028] Among them, F(d″ v ) is the vth data d″ in the current sliding window v The fusion estimation result of ν is the vth data d″ in the current sliding window v The weight of v ) is the vth data d″ in the current sliding window v The probability distribution function of .

[0029] Preferably, the S2 specifically includes:

[0030] Combined with the fusion estimation results of all data in the current sliding window, the comprehensive estimation results of the current sliding window data are calculated; based on the comprehensive estimation results of the sliding window data, the status of data collected by various IoT sensors in the city is evaluated; when the comprehensive estimation results of the sliding window data exceed the preset safety threshold, the city resource allocation and management strategy is optimized.

[0031] Preferably, the S2 specifically includes:

[0032] After obtaining the comprehensive estimation result of the current sliding window data, a new sliding window data set is generated based on the next sliding window step size, and the improved dynamic adaptive window adjustment algorithm and dynamic adaptive distribution fusion algorithm are continued to be applied to generate the comprehensive estimation result of the next sliding window data; when the sliding window reaches the end of the smooth data set and the remaining data points are not enough to fill the entire sliding window, the size of the sliding window is reduced to accommodate the remaining data points, a new sliding window data set is generated, and the dynamic adaptive distribution fusion algorithm is used to generate the comprehensive estimation result of the last sliding window data.

[0033] The beneficial effects of the technical solution of the present invention are:

[0034] 1. Through the low-pass filtering-dynamic smoothing (LPF-DS) technology, the high-frequency noise and spike outliers in the standardized sensor data set (i.e., the standardized data set) can be effectively removed. The resulting smoothed data set can more accurately reflect the actual environmental state, enhance the stability of the standardized data, and provide a reliable foundation for subsequent sliding window adjustment and data fusion.

[0035] 2. The improved dynamic adaptive window adjustment algorithm can dynamically adjust the sliding window size according to the real-time changes of the smooth data set, so that the sliding window data set can reflect the changing trend of the current data, and improve the accuracy when processing complex, nonlinear and diversified data.

[0036] 3. The dynamic adaptive distribution fusion algorithm measures the distribution dynamics of the current sliding window data to obtain the weight of the current sliding window data. Combined with the probability distribution function of the current sliding window data, the current sliding window data set is adaptively weighted fused to enhance the comprehensive processing capability of different IoT sensor data. The comprehensive estimation result of the sliding window data obtained by the fusion estimation result of all the data in the comprehensive sliding window has higher accuracy, thereby improving the practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of the method for data fusion of Internet of Things sensors for smart cities as described in the present invention. DETAILED DESCRIPTION

[0038] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0040] The specific scheme of the IoT sensor data fusion method for smart cities provided by the present invention is described in detail below with reference to the accompanying drawings.

[0041] See attached Figure 1 , which shows a flow chart of a method for data fusion of IoT sensors for smart cities provided by an embodiment of the present invention, the method comprising the following steps:

[0042] S1: Collect sensor data and perform standardization on the sensor data to obtain standardized data; obtain a standardized data set based on the standardized data; preprocess the standardized data set through low-pass filtering-dynamic smoothing technology to obtain a smoothed data set; dynamically adjust the sliding window size through an improved dynamic adaptive window adjustment algorithm, extract data from the smoothed data set based on the sliding window size, and obtain a sliding window data set;

[0043] In smart cities, many different types of IoT sensors are deployed: environmental sensors (such as temperature sensors, humidity sensors, air quality sensors), traffic sensors (such as vehicle flow counters, speed sensors) and security monitoring sensors (such as cameras, smoke detectors), etc. Sensor data is collected through IoT sensors, and the sensor data is standardized to obtain standardized data; based on the standardized data, a standardized data set is formed. The specific standardization processing method is set according to the specific implementation scenario, which will not be elaborated here.

[0044] Assume that the standardized data set expression is D = {d1, d2, ...d i , ..., d n}, where d i is the i-th standardized data; d n is the nth standardized data; i is the data index; n is the total number of standardized data.

[0045] The standardized data set is preprocessed using low-pass filtering-dynamic smoothing (LPF-DS) technology to obtain a smoothed data set. The LPF-DS technology effectively removes high-frequency noise and reduces spike data by combining low-pass filtering and dynamic smoothing strategies to eliminate noise and retain key information.

[0046] The specific implementation process of LPF-DS technology is as follows:

[0047] Firstly, a low-pass filter is applied and a filtering formula is introduced to filter the standardized data set to remove high-frequency noise and obtain a filtered data set; the filtering formula introduces Gaussian weights and logarithmic functions, which enhances the weight distribution capability on the basis of maintaining weighted average filtering, making the standardized data at the center of the filtering window more influential, and smoothing the extreme values ​​of the standardized data, so as to reduce the influence of noise on the standardized data when processing standardized data with strong random noise.

[0048] The filtering formula is as follows:

[0049]

[0050] Among them, d′ i is the i-th data in the filtered data set; d j is the jth data in the standardized data set; k is the filter window size, which is determined by the noise level of the standardized data and is set according to the specific implementation scenario; exp is an exponential function; σ is the standard deviation of the Gaussian distribution, which is used to control the decay rate of the Gaussian weight and can be determined by analyzing the noise characteristics of the standardized data and using statistical analysis or empirical formula estimation; is the Gaussian weight factor; log(1+|d j |) is used to smooth the influence of extreme values ​​of normalized data on the filtering results.

[0051] Furthermore, the dynamic smoothing formula is used to dynamically smooth the filtered data set to obtain a smoothed data set. The dynamic smoothing formula introduces logarithmic and S-shaped functions, which can capture the dynamic changes of the filtered data, smooth the spike noise, enhance the robustness to outliers, reduce the impact of edge effects on the smoothing results, and improve the stability of the smoothing results.

[0052] The dynamic smoothing formula is as follows:

[0053]

[0054] Among them, d″ i is the i-th data in the smoothed data set; α is the smoothing coefficient, which ranges from 0<α<1 and is set according to the specific implementation scenario; d″ i-1is the i-1th data in the smoothed data set. When calculating the first data in the smoothed data set, set d″0=d′1; log(1+|d′ i |) is used to amplify the i-th data d′ in the filtered data set i To ensure the stability of the smoothing process of negative filter data; is a sigmoid function.

[0055] Furthermore, an improved dynamic adaptive window adjustment algorithm is used to dynamically adjust the sliding window size, and data is extracted from the smoothed data set according to the sliding window size to obtain a sliding window data set; the improved dynamic adaptive window adjustment algorithm updates the sliding window data set by dynamically adjusting the sliding window size to ensure that the sliding window always contains smoothed data that reflects the current data change trend, and by setting a change rate threshold, a step threshold, a window size adjustment amount, and a maximum step size and a minimum step size of the sliding window, it ensures that unnecessary computing overhead is reduced when the data change is small, improves processing efficiency and accuracy, can cope with complex, nonlinear and diversified data, and provides a basis for data fusion.

[0056] The improved dynamic adaptive window adjustment algorithm process is as follows:

[0057] First, the sliding window is initialized; the initial sliding window size is set to W0, which is set by analyzing historical data or expert experience, and the historical data comes from an existing database; the first W0 smoothed data are selected in sequence from the smoothed data set to form a sliding window data set in, Represents the W0th smoothed data, which is also the W0th data in the sliding window data set. Calculate the trend value of the sliding window data set. The specific calculation formula is as follows:

[0058]

[0059] Where T is the trend value of the sliding window data set, indicating the overall change trend of the sliding window data; d″ v is the vth data in the sliding window data set; v is the index of the sliding window data; ∈ is a constant to prevent the denominator from being zero and the logarithmic calculation value from being zero, which is set according to the specific implementation scenario;

[0060] Then, the sliding window data change rate is calculated, and the sliding window size is adjusted according to the sliding window data change rate. The calculation formula for the sliding window data change rate is as follows:

[0061]

[0062] Among them, R is the sliding window data change rate, which indicates the degree of fluctuation of the sliding window data;

[0063] Next, the change rate threshold and window size adjustment amount ΔW are set by expert experience. If the sliding window data change rate R is greater than the change rate threshold, the current sliding window size is set to W=W0+ΔW; if the sliding window data change rate R is less than or equal to the change rate threshold, the current sliding window size is set to W=W0-ΔW. After dynamically adjusting the sliding window size, update the sliding window data set.

[0064] Finally, after the sliding window data set is updated, the trend value of the sliding window data set and the sliding window data change rate are recalculated, and the sliding window data change rate is compared with the change rate threshold again to determine whether the current sliding window size needs to be adjusted. The maximum number of adjustments is set according to the specific implementation scenario. When the sliding window data change rate is less than or equal to the change rate threshold, or the preset maximum number of adjustments is reached, the current sliding window data set is stopped from being updated.

[0065] Set the step threshold, maximum step and minimum step according to the specific implementation scenario. After stopping updating the current sliding window data set, compare the current sliding window data change rate with the step threshold. When the current sliding window data change rate is greater than the step threshold, set the next sliding window step to the minimum step. When the current sliding window data change rate is less than or equal to the step threshold, set the next sliding window step to the maximum step.

[0066] S2: The sliding window data set is fused through a dynamic adaptive distributed fusion algorithm to obtain a fusion estimation result of the sliding window data; the fusion estimation results of all data in the sliding window are combined to generate a comprehensive estimation result of the sliding window data.

[0067] The dynamic adaptive distributed fusion algorithm is used to fuse the current sliding window data set, which is suitable for processing complex, nonlinear and diversified data. The specific implementation process is as follows:

[0068] First, the probability distribution function of the current sliding window data is calculated using the kernel density estimation formula. The formula is as follows:

[0069]

[0070] Among them, P(d″ v ) is the vth data d″ in the current sliding window v The probability distribution function of ξ is the sum index variable, which indicates the index of the data in the current sliding window; h is the bandwidth parameter, which is determined by the cross-validation method or Silverman criterion optimization in the prior art; d″ ξ It is the ξth data in the current sliding window.

[0071] Next, calculate the distribution dynamics metric of the current sliding window data, measure the dynamic change trend of the current sliding window data, and reveal the nonlinear changes of the current sliding window data between different sliding windows. The specific calculation formula is as follows:

[0072]

[0073] in, is the distribution dynamics measure of the vth data in the current sliding window; P(d″ ξ ) is the probability distribution function of the ξ-th data in the current sliding window; is the probability distribution function of the ξth data in the previous sliding window. When calculating the distribution dynamics measure for the first time, The value is set to the probability distribution function P(d″) in the current sliding window ξ ) is calculated; when there is no ξ-th data in the current sliding window, The value is set to the probability distribution function P(d″) in the current sliding window ξ ) for calculation.

[0074] Afterwards, the weight of the current sliding window data is calculated according to the distribution dynamics measurement of the current sliding window data; based on the weight of the current sliding window data, the current sliding window data set is adaptively weighted fused to obtain the fusion estimation result of the current sliding window data; the specific calculation formula of the weight of the current sliding window data is as follows:

[0075]

[0076] Among them, w ν is the weight of the vth data in the current sliding window; is the distribution dynamics measure of the ξ-th data in the current sliding window; λ is a tuning parameter used to control the impact of the distribution dynamics measure on the weight, which is obtained through experimental tuning;

[0077] Based on the weight of the current sliding window data and the probability distribution function of the current sliding window data, the fusion estimation result of the current sliding window data is calculated. The specific formula is as follows:

[0078] F(d″ v )=w v ·P(d″ v )·log(1+|w v ·P(d″ v )|)

[0079] Among them, F(d″ v ) is the vth data d″ in the current sliding window v The fusion estimation results.

[0080] Finally, combined with the fusion estimation results of all data in the current sliding window, the comprehensive estimation result of the current sliding window data is calculated, and the formula is as follows:

[0081]

[0082] in, It is the comprehensive estimation result of the current sliding window data.

[0083] After obtaining the comprehensive estimation result of the current sliding window data, according to the next sliding window step size, the current sliding window is moved forward by the corresponding number of data points to generate a new sliding window data set, and the improved dynamic adaptive window adjustment algorithm and dynamic adaptive distribution fusion algorithm are continued to be applied to generate the comprehensive estimation result of the next sliding window data; when the sliding window reaches the end of the smooth data set and the remaining data points are not enough to fill the entire sliding window, the size of the sliding window is reduced to accommodate the remaining data points, a new sliding window data set is generated, and the dynamic adaptive distribution fusion algorithm is used to generate the comprehensive estimation result of the last sliding window data.

[0084] The comprehensive estimation result of the sliding window data is a numerical value that can be used to capture the overall change trend of the data in the sliding window. If the comprehensive estimation result of the sliding window data is large, it means that the data in the sliding window fluctuates greatly or has many outliers; if the comprehensive estimation result of the sliding window data is small, it means that the data in the sliding window is relatively stable. In smart cities, the comprehensive estimation result of the sliding window data can be used as a real-time monitoring indicator to evaluate the status of data collected by various IoT sensors in the city. Set a safety threshold according to the specific implementation scenario. When the comprehensive estimation result of the sliding window data exceeds the safety threshold, it may indicate problems such as traffic anomalies, deterioration of environmental quality, or safety hazards. Strategies can be adjusted in a timely manner, such as optimizing traffic signal control, initiating emergency response, or increasing environmental governance efforts, to optimize urban resource allocation and management strategies.

[0085] In summary, a data fusion method of IoT sensors for smart cities is completed.

[0086] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0087] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0088] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for fusion of IoT sensor data for smart cities, characterized in that: The following steps are involved: S 1: Collect sensor data and perform standardization processing on the sensor data to obtain standardized data; obtain a standardized data set based on the standardized data; preprocess the standardized data set through low-pass filtering-dynamic smoothing technology to obtain a smoothed data set; dynamically adjust the sliding window size through an improved dynamic adaptive window adjustment algorithm, extract data from the smoothed data set based on the sliding window size, and obtain a sliding window data set; S2: The sliding window data set is fused through a dynamic adaptive distribution fusion algorithm to obtain the fusion estimation result of the sliding window data; The fused estimation results of all the data in the sliding window are combined to generate a comprehensive estimation result of the sliding window data.

2. The method for fusion of IoT sensor data for smart cities according to claim 1 is characterized in that: The S1 specifically includes: In the process of implementing the low-pass filtering-dynamic smoothing technology, a low-pass filter is applied, Gaussian weights and logarithmic functions are introduced, and the standardized data set is filtered to obtain a filtered data set.

3. The method for fusion of IoT sensor data for smart cities according to claim 2 is characterized in that: The S1 specifically includes: Through the dynamic smoothing formula, the logarithmic and S-shaped functions are introduced to dynamically smooth the filtered data set to obtain a smoothed data set. The dynamic smoothing formula is as follows: Among them, d″ i is the i-th data in the smoothed data set; α is the smoothing coefficient; d′ i is the i-th data in the filtered data set; d″ i-1 is the i-1th data in the smoothed data set; is a sigmoid function.

4. The method for fusion of IoT sensor data for smart cities according to claim 3 is characterized in that: The S1 specifically includes: In the implementation process of the improved dynamic adaptive window adjustment algorithm, based on the sliding window data set, the trend value of the sliding window data set is calculated, and the calculation formula is as follows: Where T is the trend value of the sliding window data set; W0 is the initial sliding window size; d″ v is the vth data in the sliding window data set; v is the index of the sliding window data; ∈ is a constant that prevents the denominator from being zero and the logarithmic calculation value from being zero.

5. The method for fusion of IoT sensor data for smart cities according to claim 4 is characterized in that: The S1 specifically includes: Based on the trend value of the sliding window data set, the sliding window data change rate is calculated; based on the sliding window data change rate, the current sliding window size is adjusted, and the sliding window data set is updated; the sliding window data change rate is compared with a preset change rate threshold, and when the sliding window data change rate is less than or equal to the change rate threshold, or reaches a preset maximum number of adjustments, the sliding window data set is stopped from being updated; when the sliding window data set stops updating, the next sliding window step size is set based on the sliding window data change rate.

6. The method for fusion of IoT sensor data for smart cities according to claim 5 is characterized in that: The S2 specifically includes: In the process of implementing the dynamic adaptive distribution fusion algorithm, the probability distribution function of the current sliding window data is calculated by the kernel density estimation formula; based on the probability distribution function of the current sliding window data, the distribution dynamics measure of the current sliding window data is calculated.

7. The method for fusion of IoT sensor data for smart cities according to claim 6 is characterized in that: The S2 specifically includes: Based on the distribution dynamics measure of the current sliding window data, the weight of the current sliding window data is calculated.

8. The method for fusion of IoT sensor data for smart cities according to claim 7, characterized in that: The S2 specifically includes: Based on the weight of the current sliding window data and the probability distribution function of the current sliding window data, the current sliding window data set is adaptively weighted fused to obtain the fusion estimation result of the current sliding window data; the specific formula is as follows: F(d″ v )=w v ·P(d″ v )·log(1+|w v ·P(d″ v )|) Among them, F(d″ v ) is the vth data d″ in the current sliding window v The fusion estimation result of ν is the vth data d″ in the current sliding window v The weight of v ) is the vth data d″ in the current sliding window v The probability distribution function of .

9. The method for fusion of IoT sensor data for smart cities according to claim 8, characterized in that: The S2 specifically includes: Combined with the fusion estimation results of all data in the current sliding window, the comprehensive estimation results of the current sliding window data are calculated; based on the comprehensive estimation results of the sliding window data, the status of data collected by various IoT sensors in the city is evaluated; when the comprehensive estimation results of the sliding window data exceed the preset safety threshold, the city resource allocation and management strategy is optimized.

10. The method for fusion of IoT sensor data for smart cities according to claim 9, characterized in that: The S2 specifically includes: After obtaining the comprehensive estimation result of the current sliding window data, a new sliding window data set is generated based on the next sliding window step size, and the improved dynamic adaptive window adjustment algorithm and dynamic adaptive distribution fusion algorithm are continued to be applied to generate the comprehensive estimation result of the next sliding window data; when the sliding window reaches the end of the smooth data set and the remaining data points are not enough to fill the entire sliding window, the size of the sliding window is reduced to accommodate the remaining data points, a new sliding window data set is generated, and the dynamic adaptive distribution fusion algorithm is used to generate the comprehensive estimation result of the last sliding window data.

Images