Quality control method and system for generating trade video by using supply chain data AI

Abstract

The invention relates to the technical field of quality control, and discloses a quality control method and system for generating a trade video by using supply chain data AI. The method comprises the following steps: acquiring supply chain data including purchase data, production data, logistics data and sales data, extracting numerical gradient amplitude and curvature change frequency in the data to form an initial feature set, and determining a dynamic difference of the initial feature set; smoothing the supply chain data based on the dynamic difference to obtain a smoothed data sequence, calculating a gradient change curve of the smoothed data sequence through a curve algorithm, and identifying abnormal fluctuation points; generating a fusion parameter set according to the gradient change curve, inputting the fusion parameter set into a video synthesis algorithm, and generating a video frame sequence of data transition; and dynamically adjusting inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency to obtain a supply chain data stream visual video. According to the invention, the synthesis quality of the trade video is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quality control, in particular to a quality control method and system for generating trade video by using supply chain data AI. BACKGROUND

[0002] In supply chain management, the data of procurement, production, logistics and sales links has the characteristics of strong dynamics and hidden abnormal fluctuations, which makes it difficult for managers to intuitively grasp the trend and abnormal points of data flow, thereby affecting decision-making efficiency. The core technical problem is how to accurately capture abnormal fluctuations and intuitively present the trend of data flow in the business scenario of large dynamic differences and hidden abnormal fluctuations of cross-link data, to support supply chain decision-making. This problem involves multiple sub-problems: first, the time series characteristics of data in each link are complex, with significant differences in numerical gradient amplitude and curvature change frequency. How to effectively extract and integrate these features to reflect the dynamics of data is a key challenge. Second, data fluctuation abnormal points are usually hidden in a high-noise environment. How to accurately identify and mark these abnormal points through algorithms while avoiding misjudgment or omission is another difficult problem. In addition, supply chain data has strong cross-link correlation, but the non-linear change of time series makes trend analysis difficult. How to generate intuitive and dynamic time axis synchronous visualization while maintaining data smoothness is a technical difficulty. Finally, in the video synthesis process, the optimization of inter-frame smoothness and scene switching frequency needs to balance the calculation efficiency and visual effect. How to embed abnormal highlight markers in high-dynamic scenarios while ensuring the fluency and intuitiveness of the visualized video is also a problem to be solved. SUMMARY

[0003] To solve the above technical problems, the present application provides a quality control method and system for generating trade video by using supply chain data AI.

[0004] In a first aspect, the present application provides a quality control method for generating trade video by using supply chain data AI, which comprises: Step S1: acquiring supply chain data including procurement data, production data, logistics data and sales data, extracting the numerical gradient amplitude and curvature change frequency in the data to form an initial feature set, and determining the dynamic difference of the initial feature set; Step S2: performing smoothing processing on the supply chain data based on the dynamic difference to obtain a smoothed data sequence, calculating the gradient change curve of the smoothed data sequence through a curve algorithm, and identifying abnormal fluctuation points; Step S3: generating a fusion parameter set according to the gradient change curve, inputting the fusion parameter set into a video synthesis algorithm, and generating a video frame sequence of data transition; Step S4: dynamically adjusting the inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency, to obtain a supply chain data stream visualization video.

[0005] In a first implementation manner of the first aspect, the dynamic difference of the initial feature set is determined by: aligning time series of the initial feature set including the procurement data, the production data, the logistics data and the sales data by using a dynamic time warping algorithm; determining a change trend of a numerical gradient amplitude and a dynamic distribution of a curvature change frequency in the time series; calculating the dynamic difference between the procurement data, the production data, the logistics data and the sales data according to the change trend and the dynamic distribution.

[0006] In a second implementation manner of the first aspect, aligning the time series includes: establishing a cross-link time correlation matrix among the procurement, the production, the logistics and the sales, calculating a minimum warping path between the time series of different data sources, assigning a path weight based on the minimum warping path, and performing nonlinear time calibration based on the path weight.

[0007] In a third implementation manner of the first aspect, the supply chain data is smoothed to obtain a smoothed data sequence based on the dynamic difference, including: judging whether the numerical gradient amplitude in the dynamic difference exceeds a preset threshold; if the numerical gradient amplitude exceeds the preset threshold, performing a normalization processing on the supply chain data by using an adaptive interpolation method to obtain the smoothed data sequence; verifying the interval smoothness of the smoothed data sequence to obtain a smoothed data sequence meeting the smoothness requirement.

[0008] In a fourth implementation manner of the first aspect, a gradient change curve of the smoothed data sequence is calculated by a curve algorithm, and an abnormal fluctuation point is identified, including: calculating a numerical gradient amplitude of the smoothed data sequence by using a Bezier curve algorithm; determining a local extreme point according to the numerical gradient amplitude, and generating a continuous gradient change curve according to the local extreme point; calculating a second derivative of the gradient change curve as a curvature value, and defining a coordinate point corresponding to the curvature value as an abnormal fluctuation point if the curvature value is greater than a preset threshold.

[0009] In a fifth implementation form of the first aspect, the generating the fusion parameter set according to the gradient change curve comprises: analyzing a curvature change frequency of the gradient change curve to determine a trend directionality of the gradient change curve; dynamically adjusting a frame rate parameter based on the curvature change frequency; adjusting a numerical span scaling ratio according to the trend directionality; integrating the frame rate parameter and the numerical span scaling ratio to generate the fusion parameter set.

[0010] In a sixth implementation form of the first aspect, the inputting the fusion parameter set into the video synthesis algorithm to generate the video frame sequence of data transition comprises: verifying whether the frame rate parameter and the numerical span scaling ratio in the fusion parameter set are consistent with input requirements of the video synthesis algorithm, and if so, inputting the fusion parameter set into the video synthesis algorithm; after the video synthesis algorithm receives the fusion parameter set, mapping the numerical gradient amplitude in the supply chain data to a transition effect of the frame sequence to obtain a continuous and smooth video frame; calculating a length of the video frame sequence based on the frame rate parameter, and adjusting a difference value of data between frames according to the numerical span scaling ratio; integrating the curvature change frequency to generate the video frame sequence of data transition.

[0011] In a seventh implementation form of the first aspect, the inputting the fusion parameter set into the video synthesis algorithm to generate the video frame sequence of data transition comprises: extracting a numerical gradient amplitude and a local extreme point of each video frame from the video frame sequence; mapping the numerical gradient amplitude to a color tone by using a color mapping function; generating a video frame sequence after color configuration based on the adjustment of the mapping according to the numerical span scaling ratio; synchronizing a time axis of the video frame sequence to obtain a video frame sequence containing time axis synchronization.

[0012] In an eighth implementation form of the first aspect, the dynamically adjusting the inter-frame transition smoothness and the scene switching frequency of the video frame sequence based on the curvature change frequency comprises:

[0013] calculating a numerical gradient amplitude difference of each frame based on the time axis synchronization, and applying a smoothing filter to each frame based on the difference value for smoothing processing; analyzing a local extreme point in the smoothed data sequence to determine a curvature change frequency threshold of a transition region, and adjusting a pixel fusion ratio of each frame according to a threshold result to realize smoothness optimization. calculating an average curvature change frequency of the gradient change curve, and adjusting a scene switching frequency of the video frame sequence based on the average curvature change frequency; embedding a highlight mark of the abnormal fluctuation point in the optimized video frame sequence frame by frame to obtain the supply chain data flow visualization video.

[0014] In a second aspect, the application provides a quality control system for generating trade videos using supply chain data AI, which comprises: An acquisition module acquires supply chain data including procurement data, production data, logistics data, and sales data, extracts numerical gradient amplitude and curvature change frequency in the data to form an initial feature set, and determines the dynamic difference of the initial feature set; An anomaly recognition module performs smoothing processing on the supply chain data based on the dynamic difference to obtain a smoothed data sequence, calculates a gradient change curve of the smoothed data sequence through a curve algorithm, and identifies abnormal fluctuation points; A synthesis module generates a fusion parameter set according to the gradient change curve, inputs the fusion parameter set into a video synthesis algorithm, and generates a video frame sequence of data transition; An adjustment module dynamically adjusts the inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency to obtain the supply chain data flow visualization video.

[0015] Compared with the prior art, the application has at least the following advantages: The quality control method for generating trade videos using supply chain data AI provided by the application can efficiently and accurately align, smooth, and optimize data at each link of the supply chain (such as procurement, production, logistics, and sales) by comprehensively applying dynamic time warping algorithm, Bezier curve algorithm, numerical gradient amplitude, curvature change frequency, and other technologies, thereby generating a supply chain data flow visualization video with high visibility and accuracy. Specifically, the method can automatically identify and process abnormal fluctuation points in the data, enhance the intuitiveness of data display by using color mapping and numerical span scaling technology, and help users quickly identify potential problems. In addition, by dynamically adjusting the smoothness and scene switching frequency of the video frame sequence, the smoothness of the video when displaying data fluctuations is ensured, the abruptness of data display is reduced, and the viewing experience of the video is improved. Finally, the abnormal fluctuation points embedded with highlight marks effectively highlight the key changes in the data flow, ensuring that users can focus on important information. This method not only optimizes the display effect of supply chain data, but also improves the accuracy and practicality of data analysis, greatly enhancing the support function of data-driven decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0017] Figure 1 An embodiment schematic diagram of the quality control method for generating trade video by AI using supply chain data in the embodiments of the present application; Figure 2 A synthesis flow chart of the video frame sequence in the embodiments of the present application; Figure 3 A video quality evaluation comparison chart of the method of the present application and the traditional method in the embodiments of the present application; Figure 4 An embodiment schematic diagram of the quality control system for generating trade video by AI using supply chain data in the embodiments of the present application. DETAILED DESCRIPTION

[0018] The embodiments of the present application provide a quality control method and system for generating trade video by AI using supply chain data. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "comprising" or "having" and any variation thereof is intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0019] For the convenience of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 An embodiment of the quality control method for generating trade video by AI using supply chain data in the embodiments of the present application includes: Step S1, acquiring supply chain data including procurement data, production data, logistics data and sales data, extracting the value gradient amplitude and curvature change frequency in the data to form an initial feature set, and determining the dynamic difference of the initial feature set.

[0020] Specifically, in modern supply chain management, the data of each link such as procurement, production, logistics and sales fluctuates greatly, and these fluctuations are usually affected by many factors, such as changes in market demand, production cycle, logistics delay, etc. The trend of these data changes is often difficult to intuitively grasp, and abnormal fluctuations may have a serious impact on decision-making. In order to identify potential problems in dynamic and complex supply chain data, the data of each link must first be analyzed in detail and key features must be extracted. Real-time procurement data, production data, logistics data and sales data are collected from each link of the supply chain. These supply chain data are usually time series data, reflecting the operation of each link. For example, procurement data may reflect the procurement quantity of an item within a certain period of time, production data includes the production quantity within a certain period of time, logistics data records the transportation time and distance, and sales data records the sales amount or product sales, etc.

[0021] The numerical gradient amplitude and curvature change frequency in the time series are extracted, wherein the numerical gradient amplitude reflects the change rate between the data points of the time series, helping to identify abnormal fluctuations existing in some links; the curvature change frequency reflects the acceleration of the change of the time series data, that is, the frequency of data fluctuations. The greater the curvature, the faster the data changes, and the higher the frequency. By analyzing the numerical gradient amplitude and the curvature change frequency, the dynamic differences between different links (procurement, production, logistics and sales) are calculated, and their differences in fluctuations, trends and accelerations are measured. The Euclidean distance is usually used to measure the differences between these data. The dynamic difference reflects the degree of data change at different links or different time points. The DTW algorithm can effectively align the time axes of the two, eliminate the inconsistency in time, and thus make the subsequent analysis more accurate. The specific calculation process is described later.

[0022] Step S2, based on the dynamic difference, the supply chain data is smoothed to obtain a smoothed data sequence. The gradient change curve of the smoothed data sequence is calculated by a curve algorithm, and the abnormal fluctuation points are identified.

[0023] Specifically, in order to make the data more easily analyzed and reveal its true trend of change, it is necessary to smooth the data, identify the rate of change of the data by calculating the gradient amplitude of the data sequence, and if the gradient amplitude of some data points is too large, it means that these data points have strong fluctuations and may contain noise or anomalies, so before smoothing, it is necessary to first judge these abnormal large value gradient amplitudes, and when the value gradient amplitude exceeds the preset threshold, the adaptive interpolation method is used to normalize the data. Then use a curve algorithm such as the Bezier curve algorithm to calculate the gradient change of the smoothed data sequence. The Bezier curve can effectively fit the data points and smooth the fluctuations of the data, generating a continuous and smooth gradient change curve, identifying abnormal fluctuation points on the gradient change curve, and the specific identification process will be described later. Through the above steps, the supply chain data is effectively smoothed to eliminate short-term fluctuations and noise, and the gradient change curve is calculated by the curve algorithm to help identify abnormal fluctuation points, providing more stable and accurate input for subsequent data analysis and decision-making.

[0024] Step S3, generating a fusion parameter set according to the gradient change curve, inputting the fusion parameter set into the video synthesis algorithm, and generating a video frame sequence of data transition.

[0025] Specifically, in order to ensure that the data flow shown in the video is natural and consistent with the actual business scenario, it is necessary to consider data fluctuations, smoothness and visual effects through the gradient change curve, and then generate a suitable fusion parameter set. The fusion parameter set includes the frame rate parameter of the video frame and the numerical span scaling ratio, wherein the frame rate parameter determines the number of frames generated per second to ensure the smoothness of the video, and the numerical span scaling ratio ensures that the visual effect of data change adapts to the change amplitude of the data. The specific generation method will be described in detail later. According to the input frame rate parameter and numerical span scaling ratio, the video synthesis algorithm smoothly transitions between data changes, ensuring that there are no abrupt jumps or stalls when playing the video. According to the fusion parameter set generated by the gradient change curve, the video synthesis algorithm can generate a high-quality video frame sequence that conforms to the data fluctuation characteristics. The final data transition video can accurately present the change of the supply chain data flow, helping managers to more intuitively analyze and make decisions.

[0026] Step S4, dynamically adjusting the inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency, and obtaining a supply chain data flow visualization video.

[0027] Specifically, if the inter-frame transition smoothness of the video is poor, it may cause the picture to freeze or the data change to be unclear, and if the scene switching frequency is too high or too low, it may affect the visual effect of the video, making it lack of smoothness and rhythm. In order to solve these problems, the application proposes to dynamically adjust the inter-frame transition smoothness and scene switching frequency of the video frame sequence through the curvature change frequency. Specifically, by calculating the curvature change frequency, it is identified which part of the data fluctuation is more violent, and then the detailed display of the video playback is adjusted according to the data change characteristics, to ensure that the finally generated visual video is both smooth and expressive. Through the above method, the effect of video playback can be fine-tuned according to the change characteristics of the data, for example, the high fluctuation data segment gets more frame number and scene switching frequency, so that it is more prominent in the video, and the low fluctuation data segment saves computing resources by reducing the switching frequency and frame number, while ensuring the smoothness and clarity of the video.

[0028] In a specific embodiment, determining the dynamic difference of the initial feature set specifically includes the following steps: The time series of the procurement data, production data, logistics data and sales data in the initial feature set are analyzed using the dynamic time warping algorithm, and all time series are aligned; The change trend of the numerical gradient amplitude and the dynamic distribution of the curvature change frequency in the time series are determined; According to the change trend and the dynamic distribution, the dynamic difference between the procurement data, production data, logistics data and sales data is calculated.

[0029] Specifically, the time series of the procurement data, production data, logistics data and sales data in the initial feature set are analyzed using the dynamic time warping algorithm, wherein the dynamic time warping algorithm aligns non-equal length sequences by calculating the shortest path distance between time series, ensures that the data of different links of the supply chain is compared synchronously in time dimension, according to the numerical gradient amplitude and the curvature change frequency extracted from the initial feature set, the time series of the procurement data is constructed as a reference sequence, and the time series of the production, logistics and sales data are aligned with it, the distance matrix after alignment is calculated, wherein the elements of the distance matrix represent the Euclidean distance between sequence points, the optimal warping path of the distance matrix is solved by dynamic programming, and the aligned time series is obtained.

[0030] After aligning the time series data, the change trend of the numerical gradient amplitude and the dynamic distribution of the curvature change frequency in the time series are determined, wherein the gradient amplitude sequence of the time series is calculated by difference, and the trend line is fitted to identify the rising or falling mode, the curvature change frequency refers to the number of times the curvature of the sequence changes in a local interval, the curvature value is calculated by second difference, and the change frequency in unit time is counted to form a distribution graph, for example, the time series data of a certain link is , , ,......, wherein the gradient amplitude is calculated by the formula: , and the curvature change frequency is calculated by the formula: , the curvature is calculated by applying the finite difference method to the aligned time series, the curvature is defined as the approximate value of the second derivative of the sequence function, the change points with curvature values exceeding the preset threshold are counted to generate a dynamic distribution; according to the trend and the dynamic distribution, the dynamic difference between the procurement data, the production data, the logistics data and the sales data is calculated, wherein the trend line slope and the distribution variance are taken as inputs, the Euclidean distance is taken as a difference measure, and the gradient amplitudes of the sales data are assumed to be 5, 5 and 5, and the gradient amplitudes of the production data are assumed to be 10, 10 and 10, and the dynamic difference calculation formula is: , the dynamic difference between the sales data and the production data is 8.66, which is obtained by substituting the calculation, and the above method can effectively handle the time offset and volatility of the supply chain data, and help accurately understand the correlation between different links.

[0031] In a specific embodiment, aligning all time series specifically includes the following steps: A cross-link time correlation matrix between procurement, production, logistics and sales is established, the minimum regular path between sequences is calculated to align the time series of different data sources, and the path weight is given based on the minimum regular path, and the nonlinear time calibration is performed based on the path weight.

[0032] Specifically, since there is a time difference between each link (procurement, production, logistics and sales) in the supply chain, and the data collection time points of each link are different, direct comparison of them may not obtain accurate results, therefore, the time series of different data sources need to be aligned through the minimum regular path to ensure that they can be compared synchronously in the time dimension, for example, assuming that there is a procurement data sequence: A1=[100, 110, 120, 130] (unit: pieces), and a production data sequence: B1=[105, 115, 125, 140, 150] (unit: pieces), wherein the procurement data has 4 data points and the production data has 5 data points, the Euclidean distance between each data point in the procurement data sequence and the production data sequence is calculated to obtain a distance matrix, each element of the matrix represents the distance between two time points, the shortest path is found by calculating the distance matrix and using dynamic programming, in order to further refine the alignment result, a path weight is given to each part of the path, and the weight reflects the reliability of the data alignment of the part, for example, a smaller distance corresponds to a larger weight, which means that the quality of the data alignment of the part is higher; the time is nonlinearly calibrated according to the path weight, the time step is adjusted to make different parts of the time series transition more smoothly, and the error caused by the time difference is reduced.

[0033] In a specific embodiment, the smoothing of the supply chain data based on the dynamic difference includes the following steps: determining whether the gradient amplitude of the dynamic difference exceeds a preset threshold value; if the gradient amplitude exceeds the preset threshold value, using an adaptive interpolation method to normalize the supply chain data to obtain a smoothed data sequence; verifying the interval smoothness of the smoothed data sequence to obtain a smoothed data sequence that meets the smoothness requirement.

[0034] Specifically, the calculated gradient amplitude is compared with the preset threshold value, and the threshold value is set based on the fluctuation range of historical supply chain data. For example, the threshold value is 10%. For example, in the context of supply chain procurement data, if the gradient amplitude represents a change of 50% from 100 units to 150 units, which exceeds the threshold value of 10%, the adaptive interpolation method is used to normalize the data stream. Specifically, it includes first determining the data stream segment corresponding to the gradient amplitude that exceeds the threshold value, which is the procurement or logistics time series extracted from the initial feature set. Secondly, the adaptive interpolation method is selected to dynamically adjust the interpolation points according to the density of the data points. For example, linear interpolation is used in combination with local density weight to reduce interpolation points when data points are dense to avoid overfitting. Then, the data stream is normalized, i.e. the value is mapped to the interval [0, 1], which is processed by the min-max normalization formula, for example, the value x is normalized as (x-min) / (max-min). Combining interpolation to fill in missing or abnormal points, iteratively adjusting interpolation parameters until the gradient amplitude is reduced to within the threshold value, forming a closed loop processing to ensure data stream smoothing.

[0035] In a specific embodiment, the gradient change curve of the smoothed data sequence is calculated by a curve algorithm, and the abnormal fluctuation point is identified, including the following steps: using a Bezier curve algorithm to calculate the numerical gradient amplitude of the smoothed data sequence; determining the local extreme points according to the numerical gradient amplitude, and generating a continuous gradient change curve according to the local extreme points; calculating the second derivative of the gradient change curve as the curvature value, and if the curvature value is greater than a preset threshold value, defining the coordinate point corresponding to the curvature value as an abnormal fluctuation point.

[0036] Specifically, the Bezier curve algorithm is a parameterization method for generating smooth curves based on control points, which is used to fit the point set in the smoothed data sequence in this paper, and the slope change between adjacent points is calculated as the numerical gradient amplitude; for example, for the purchase data sequence in the supply chain, first select the key data points in the sequence as the control points of the Bezier curve, then calculate the derivative at each point through the curve equation to get the gradient amplitude, and extract the control point set according to the smoothed data sequence, wherein the control points correspond to the numerical peak and valley values of the purchase, production, logistics or sales data, and the control point set is interpolated using the Bezier curve formula to generate a smooth curve, wherein the curve parameter t changes from 0 to 1; the tangent vector of the curve at each t is calculated as the gradient direction, and the tangent vector is taken as the modulus to get the numerical gradient amplitude sequence, and by scanning the numerical gradient amplitude sequence, the turning point of the gradient sign from positive to negative or negative to positive is identified as a local extreme point; for example, in the sales data fluctuation, the point whose gradient amplitude exceeds the average value is marked as a potential extreme value; the local extreme points are used as anchor points to connect these points to form a continuous gradient change curve, and the position of the abnormal fluctuation point is analyzed through the gradient change curve, and the position of the abnormal fluctuation point is analyzed by calculating the point whose curvature exceeds the preset threshold. By calculating the gradient change curve of the smoothed data through the Bezier curve algorithm and identifying the local extreme points, the fluctuation trend of the data can be effectively fitted, and the potential abnormal fluctuation points can be accurately located.

[0037] In a specific embodiment, generating the fusion parameter set according to the gradient change curve specifically includes the following steps: analyzing the curvature change frequency of the gradient change curve to determine the trend directionality of the gradient change curve; adjusting the frame rate parameter dynamically based on the curvature change frequency; adjusting the numerical span scaling ratio according to the trend directionality; integrating the frame rate parameter and the numerical span scaling ratio to generate the fusion parameter set.

[0038] Specifically, the curvature change frequency of the gradient change curve is analyzed, where the curvature change frequency is obtained by calculating the second derivative change rate of the curve at consecutive points, the trend directionality of the gradient change curve is determined, the trend directionality is determined by fitting the first derivative sign sequence of the curve, if most of the derivatives are positive, the trend is rising, otherwise it is falling or stable; if the curvature change frequency is higher than the preset upper limit, the frame rate dynamic adjustment parameter is set to a higher value to capture rapid changes, if the curvature change frequency is lower than the preset lower limit, the frame rate dynamic adjustment parameter is set to a lower value to smooth the display; for example, in the analysis of procurement data, high-frequency curvature change represents a sudden surge in procurement volume, by increasing the frame rate, the video can display these changes more finely, thereby improving the dynamic interactivity of the visual video; if the trend directionality is rising, the numerical span scaling ratio is increased to expand the display range, if the trend directionality is falling, the ratio is reduced to compress the range, for example, in the scenario of sales data fluctuation, if the trend directionality is rising, the numerical span scaling ratio is set to 1.5 times, which makes the video synthesis algorithm process a larger numerical range, generate a video frame sequence containing more details, which is beneficial to analyze the span distribution characteristics and obtain an optimized data flow feature set, integrate the frame rate dynamic adjustment parameter and the numerical span scaling ratio, generate a fusion parameter set, which is used as the input of the subsequent video synthesis algorithm.

[0039] In a specific embodiment, inputting the fusion parameter set into the video synthesis algorithm to generate a video frame sequence of data transition specifically includes the following steps: Verify whether the frame rate parameter and the numerical span scaling ratio in the fusion parameter set are consistent with the input requirements of the video synthesis algorithm, if consistent, input the fusion parameter set into the video synthesis algorithm; After the video synthesis algorithm receives the fusion parameter set, the numerical gradient amplitude in the supply chain data is mapped to the transition effect of the frame sequence to obtain continuous and smooth video frames; Based on the frame rate parameter, the length of the video frame sequence is calculated, and the difference value of the inter-frame data is adjusted according to the numerical span scaling ratio; Integrate the curvature change frequency to generate a video frame sequence of data transition.

[0040] Specifically, as Figure 2As shown, it is a synthetic flow chart of video frame sequence, according to the input specification predefined by the video synthesis algorithm, the frame rate dynamic adjustment parameter in the fusion parameter set is extracted, which represents the dynamic rate change of video frame generation, the extracted parameter is compared with the frame rate range required by the algorithm, for example, the required range is 24 to 60 frames per second, if the frame rate dynamic adjustment parameter of the fusion parameter set falls within this range, it is marked as preliminary consistent; the numerical span scaling ratio is extracted, which defines the mapping scaling of data value to video frame pixel, used to process the amplitude change of supply chain numerical gradient, check whether the ratio meets the scaling threshold of the algorithm, for example, the threshold is 0.5 to 2.0, if it meets, confirm the overall consistency; after verification, the fusion parameter set is directly transmitted to the interface of the video synthesis algorithm as input data, ensuring that parameters such as frame rate dynamic adjustment parameters and numerical span scaling ratios are read by the algorithm, after receiving the fusion parameter set, the video synthesis algorithm processes the supply chain data sequence based on the variant of dynamic time warping algorithm, for example, the numerical gradient amplitude of procurement data is mapped to the transition effect of frame sequence, to generate continuous frames, calculate the length of frame sequence, adjust the data difference between frames according to the numerical span scaling ratio, for example, if the ratio is 1.5, the gradient change is amplified to produce more obvious transition frames, integrate the curvature change frequency, generate video frame sequence; for example, in production data processing, if the numerical gradient amplitude rises from 10 to 50, the algorithm generates a transition frame sequence to show the gradual change process, frame 1 shows low-intensity transition corresponding to amplitude 10, frame 50 shows high-intensity transition of amplitude 50, which is beneficial to users to quickly identify fluctuation trends.

[0041] In a specific embodiment, inputting the fusion parameter set into the video synthesis algorithm to generate a video frame sequence of data transition specifically includes the following steps: Extracting the numerical gradient amplitude and local extreme points of each video frame from the video frame sequence; Mapping the numerical gradient amplitude to the hue using a color mapping function; Adjusting the mapping based on the numerical span scaling ratio to generate a video frame sequence with color configuration; Synchronizing the time axis of the video frame sequence to obtain a video frame sequence containing time axis synchronization.

[0042] Specifically, the display of data fluctuations not only involves simple numerical changes, but also needs to enhance the visual effect through color and transition effects. In order to ensure the clarity, smoothness and information transmission effect of the video, accurate color mapping and numerical adjustment of the video frames are needed, while ensuring the time synchronization of the data in each link. The numerical gradient amplitude and local extreme points of each frame are extracted from the video frame sequence, and a color mapping function such as the HSV color model is used to map the amplitude to the hue, for example, low amplitude is mapped to blue and high amplitude is mapped to red, to ensure that the color reflects the data intensity. The scaling ratio of the adjustment mapping is based on the numerical span, for example, a linear mapping is maintained when the ratio is 1.0, and the contrast is enhanced when the ratio is 1.5, so that the abnormal fluctuation points are highlighted in bright yellow. Apply to the entire sequence to form a frame sequence after color configuration, where the color change is synchronized with the data transition. For example, in the supply chain data flow, the low gradient amplitude of the procurement data is configured as light blue and the peak value of the production data is configured as dark red. This configuration can intuitively distinguish the fluctuation intensity of different links and is beneficial to the rapid positioning of problem areas by analysts. In addition, the timestamp of each frame in the video frame sequence is identified, the time axis is aligned based on the initial time sequence of the supply chain data flow using the linear interpolation method, and the time interval of the frame sequence is matched with the actual length of the data flow. For example, if the data span is 1 hour, the total length of the frame sequence is synchronized to the video length of 60 seconds. If the deviation is less than 0.1 seconds, it is marked as synchronization completed, otherwise the frame rate is adjusted dynamically to adjust the parameters and realign. The final video frame sequence is generated, and the time axis label is embedded, such as the actual data time point is marked every 10 frames, to ensure synchronization. Through color mapping and numerical span scaling, the visual performance of data transition is more accurate and intuitive, and the time axis synchronization ensures that the data in different links can be effectively aligned in the video, providing a supply chain visualization display with time consistency and data visibility.

[0043] As Figure 3As shown, it is a video quality evaluation comparison chart of the method of the present application and the traditional method, which shows the comparison of the method of the present application and the traditional method in four main video quality evaluation indexes. The evaluation indexes include: transition smoothness, scene switching frequency, abnormal fluctuation point recognition rate and frame rate smoothness. The data in the chart reflects the performance of the two methods under each index. The horizontal coordinate represents the evaluation index, and the vertical coordinate represents the score of the method under these indexes. The video synthesis method in the prior art usually refers to the traditional data visualization-based method, which directly converts data into video display, which is represented by a darker gray line. The data fluctuates greatly, the transition is not smooth enough, and there may be abrupt changes between some video frames. The traditional method is mainly based on raw data to generate video content, and lacks intelligent data processing and optimization, so the score of the traditional method is low. The present application is represented by a line close to black, showing smooth transition effect, and the data changes continuously and naturally. The present application calculates the numerical gradient amplitude and curvature change frequency of the data, and performs smoothing processing on the video frames according to the fluctuation intensity of the data, reduces the abrupt fluctuation, and ensures the smooth transition of data display.

[0044] In a specific embodiment, dynamically adjusting the inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency specifically includes the following steps: Calculate the numerical gradient amplitude difference of each frame based on the time axis synchronization, and apply a smoothing filter for smoothing processing based on the difference value frame by frame; Analyze the local extreme points in the smoothed data sequence, determine the curvature change frequency threshold of the transition region, and adjust the pixel fusion ratio frame by frame according to the threshold result to realize the smoothness optimization; Calculate the average curvature change frequency of the gradient change curve, and adjust the scene switching frequency of the video frame sequence based on the average curvature change frequency; Embed the highlight mark of the abnormal fluctuation point in the optimized video frame sequence frame by frame to obtain the supply chain data flow visualization video.

[0045] Specifically, the inter-frame smoothing algorithm is used to process the video frame sequence, which smoothes the transition effect of the data by calculating the pixel difference of adjacent video frames, reduces the data fluctuation by identifying the edge contours of adjacent frames and fusing these contours using a weighted average method, ensures the natural and smooth transition of the video, calculates the numerical gradient amplitude difference of each frame based on the time axis synchronization of the video frame sequence as the input data of the smoothing process, which helps to detect the transition points between video frames and reduce the impact of abrupt points in the smoothing process; according to the local extreme points of the smoothed data sequence, the curvature change frequency of the transition region is calculated, and the smoothing threshold is dynamically adjusted to optimize the video transition. Assuming that the sales data fluctuates greatly, the curvature value of a certain frame in the calculated gradient amplitude sequence is 0.6, which exceeds the threshold value 0.5, so the transition of this frame is more smoothly processed, making the transition more smooth. Based on the frame rate dynamic adjustment parameter in the fusion parameter set, the scene switching frequency is adjusted, and by analyzing the curvature change frequency and trend directionality, the switching frequency is dynamically modified to adapt to the fluctuation of the data stream, for example, when the data fluctuation frequency is high, the scene switching frequency is reduced to avoid information overload. For a scene with large sales data fluctuation, if the average curvature frequency exceeds 1.2, the scene switching frequency can be adjusted to 2 times per second to avoid viewing fatigue caused by too much information, while for smooth production data, the scene switching frequency is maintained at 4 times per second; according to the calculation result, the linear scaling method is applied to adjust the switching rhythm of the video sequence, so as to realize the smooth transition of the video stream, and by adjusting the switching speed between different scenes, the display of the data stream is more natural and smooth. By color mapping technology, abnormal fluctuation points are marked as red highlights to enhance their visibility in the video. Assuming that the curvature change frequency of a certain fluctuation point in the sales data is abnormally high, its synchronous position on the time axis is calculated, and then the point is marked as a red highlight to highlight in the video, making it easy to identify. From the continuous gradient change curve, locate the abnormal fluctuation point, analyze the numerical gradient amplitude and curvature to determine which points are abnormal fluctuations, integrate the results of the foregoing optimization and highlight marking, and directly generate a video output containing all the highlight markers and smooth transitions. The final video not only clearly shows the data fluctuation trend, but also highlights the abnormal fluctuation points, improving the accuracy of data analysis. The above optimization process effectively improves the viewing experience of the video, making the trend, fluctuation and abnormal points of the data more clearly visible, providing a more efficient analysis tool for supply chain management decision-making.

[0046] The above describes a quality control method for generating trade videos using supply chain data AI in an embodiment of the present application. The quality control system for generating trade videos using supply chain data AI in an embodiment of the present application is described below. Please refer to Figure 4 The quality control system for generating trade videos using supply chain data AI in an embodiment of the present application includes: An acquisition module acquires supply chain data including procurement data, production data, logistics data, and sales data, extracts a value gradient amplitude and a curvature change frequency in the data to form an initial feature set, and determines a dynamic difference of the initial feature set; An anomaly identification module performs smoothing processing on the supply chain data based on the dynamic difference to obtain a smoothed data sequence, calculates a gradient change curve of the smoothed data sequence through a curve algorithm, and identifies an abnormal fluctuation point; A synthesis module generates a fusion parameter set according to the gradient change curve, inputs the fusion parameter set into a video synthesis algorithm, and generates a video frame sequence of data transition; An adjustment module dynamically adjusts interframe transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency, and obtains a supply chain data stream visualization video.

[0047] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, system and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0048] The integrated unit, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0049] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A quality control method for AI-generated trade videos using supply chain data, the method comprising: The method comprises: Step S1: acquiring supply chain data including procurement data, production data, logistics data and sales data, extracting value gradient amplitude and curvature change frequency in the data to form an initial feature set, and determining dynamic differences of the initial feature set; Step S2: performing smoothing processing on the supply chain data based on the dynamic differences to obtain a smoothed data sequence, calculating a gradient change curve of the smoothed data sequence through a curve algorithm, and identifying abnormal fluctuation points; Step S3: generating a fusion parameter set according to the gradient change curve, inputting the fusion parameter set into a video synthesis algorithm to generate a video frame sequence of data transition; Step S4: dynamically adjusting inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency to obtain a supply chain data stream visualization video.

2. The method of claim 1, wherein, Determining the dynamic differences of the initial feature set comprises: analyzing time series of the procurement data, production data, logistics data and sales data in the initial feature set by using a dynamic time warping algorithm, aligning all the time series; determining change trends of value gradient amplitudes and dynamic distributions of curvature change frequencies in the time series; calculating dynamic differences between the procurement data, production data, logistics data and sales data according to the change trends and dynamic distributions.

3. The method of claim 2, wherein, Aligning all the time series comprises: establishing a cross-link time correlation matrix among procurement, production, logistics and sales, calculating a minimum warping path alignment between sequences to align time series of different data sources, assigning path weights based on the minimum warping path, and performing nonlinear time calibration based on the path weights.

4. The method of claim 1, wherein, Performing smoothing processing on the supply chain data based on the dynamic differences to obtain a smoothed data sequence comprises: judging whether the value gradient amplitude in the dynamic differences exceeds a preset threshold; if the value gradient amplitude exceeds the preset threshold, performing normalization processing on the supply chain data by using an adaptive interpolation method to obtain a smoothed data sequence; verifying interval smoothness of the smoothed data sequence to obtain a smoothed data sequence meeting smoothness requirements.

5. The method of claim 1, wherein, Calculating a gradient change curve of the smoothed data sequence through a curve algorithm and identifying abnormal fluctuation points comprises: calculating a value gradient amplitude of the smoothed data sequence by using a Bezier curve algorithm; determining local extreme points according to the value gradient amplitude, and generating a continuous gradient change curve according to the local extreme points; calculating a second derivative of the gradient change curve as a curvature value, and defining a coordinate point corresponding to the curvature value as an abnormal fluctuation point if the curvature value is greater than a preset threshold.

6. The method of claim 1, wherein, Generating a fusion parameter set according to the gradient change curve comprises: analyzing curvature change frequency of the gradient change curve, determining trend directionality of the gradient change curve; dynamically adjusting a frame rate parameter based on the curvature change frequency; adjusting a value span scaling ratio according to the trend directionality; integrating the frame rate parameter and the value span scaling ratio to generate the fusion parameter set.

7. The method of claim 1, wherein, Inputting the fusion parameter set into a video synthesis algorithm to generate a video frame sequence of data transition comprises: verify whether the frame rate parameter and the numerical span scaling ratio in the fusion parameter set are consistent with the input requirements of the video synthesis algorithm, and if so, input the fusion parameter set into the video synthesis algorithm; After the video synthesis algorithm receives the fusion parameter set, the numerical gradient amplitude in the supply chain data is mapped into the transition effect of the frame sequence to obtain a continuous and smooth video frame; Based on the frame rate parameter, the length of the video frame sequence is calculated, and the difference value of the inter-frame data is adjusted according to the numerical span scaling ratio; Integrate the curvature change frequency to generate a video frame sequence of data transition.

8. The method of claim 7, wherein, Inputting the fusion parameter set into the video synthesis algorithm to generate a video frame sequence of data transition further includes: Extracting the numerical gradient amplitude and local extreme points of each video frame from the video frame sequence; Mapping the numerical gradient amplitude to the hue using a color mapping function; Adjust the mapping based on the numerical span scaling ratio to generate a video frame sequence with color configuration; Synchronize the time axis of the video frame sequence to obtain a video frame sequence containing time axis synchronization.

9. The method of claim 1, dynamically adjusting the inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency, comprising: Based on the time axis synchronization, calculate the difference of the numerical gradient amplitude of each frame, and apply a smoothing filter frame by frame based on the difference value for smoothing processing; Analyze the local extreme points in the smoothed data sequence, determine the curvature change frequency threshold of the transition region, and adjust the pixel fusion ratio frame by frame according to the threshold result to realize smoothness optimization; Calculate the average curvature change frequency of the gradient change curve, and adjust the scene switching frequency of the video frame sequence based on the average curvature change frequency; Embed highlight markers of abnormal fluctuation points in the optimized video frame sequence frame by frame to obtain a supply chain data flow visualization video.

10. A quality control system for AI-generated trade videos using supply chain data, for implementing a method for AI-generated trade videos using supply chain data according to any one of claims 1-8, characterized in that, The system comprises: An acquisition module acquires supply chain data including procurement data, production data, logistics data, and sales data, extracts the numerical gradient amplitude and curvature change frequency in the data to form an initial feature set, and determines the dynamic difference of the initial feature set; An anomaly identification module performs smoothing processing on the supply chain data based on the dynamic difference to obtain a smoothed data sequence, calculates the gradient change curve of the smoothed data sequence through a curve algorithm, and identifies abnormal fluctuation points; A synthesis module generates a fusion parameter set according to the gradient change curve, inputs the fusion parameter set into a video synthesis algorithm, and generates a video frame sequence of data transition; An adjustment module dynamically adjusts the inter-frame transition smoothness and scene switching frequency of the video frame sequence based on the curvature change frequency to obtain a supply chain data flow visualization video.

Images