A dynamic data evaluation system and method for brand value of agricultural products

Abstract

The application relates to the technical field of data processing and digital evaluation, and discloses a dynamic data evaluation system and method for brand value of agricultural products, which comprises the following steps: a target and background double monitoring domain is established by using a heterogeneous data acquisition and alignment module; unstructured marketing text is converted into quantitative physical commitment features through a semantic physical vector mapping engine; a double-track entropy value calculation module is used to calculate brand consistency entropy representing brand performance stability and environment background entropy representing average dispersion of the industry based on Monte Carlo simulation and probability density estimation; an entropy difference value evaluation module is used to calculate net efficiency according to the difference between the two and an environmental pressure compensation factor, and a dynamic value index is generated by combining a time series accumulation model; and a dynamic updating terminal is used to smooth the index and construct a value account book. Through physical measurement and information entropy difference mechanism, the application effectively eliminates environmental interference, realizes objective quantification and real-time dynamic monitoring of brand value.

Description

Technical Field

[0001] This invention relates to the field of data processing and digital evaluation technology, specifically to a dynamic data evaluation system and method for the brand value of agricultural products. Background Technology

[0002] The brand value of agricultural products not only reflects the product's market premium capability, but also comprehensively embodies consumers' trust in the consistency of product quality and the reliability of the supply chain. With the development of agricultural IoT and digital supply chain technologies, although the production and circulation process of agricultural products generates big data on the physical state, existing brand evaluation systems still mainly follow the traditional intangible asset evaluation logic.

[0003] Current methods for evaluating agricultural product brands typically rely on financial benefit analysis, consumer surveys, or expert scoring models. This evaluation model, based on subjective perception or historical financial data, suffers from significant lag and subjective bias. Existing evaluation data sources are mostly static statistical reports or subjective evaluations, lacking real-time correlation with the microscopic physical conditions during the agricultural product distribution process. High-quality promises such as end-to-end cold chain logistics and guaranteed sugar content issued by the marketing end are often disconnected from sensor data on temperature, vibration, and physicochemical indicators generated at the logistics end. There is a lack of technical means to semantically align and verify unstructured marketing text with structured physical measurement data, resulting in brand value evaluation results that fail to accurately reflect the actual physical fulfillment level of the product.

[0004] Agricultural products are highly perishable, and their quality is easily affected by external objective conditions such as the natural environment and logistics infrastructure. Existing assessment models typically employ absolute value evaluation methods, failing to establish an effective environmental benchmark mechanism. Under harsh climatic or logistical conditions, the management costs and technological efforts brands make to maintain quality are often overlooked; conversely, in favorable natural environments, even low-level management can achieve high quality scores. This assessment method, lacking environmental benchmarks, cannot distinguish whether quality fluctuations stem from force majeure environmental factors or the brand's own management capabilities, resulting in assessment results that cannot accurately quantify the brand's core competitiveness against environmental entropy increases. Furthermore, traditional assessments are mostly periodic and static evaluations, unable to form continuous time-series tracking, and thus failing to meet the needs of real-time dynamic monitoring of brand reputation in the context of the digital economy. Summary of the Invention

[0005] The first aspect of this invention provides a dynamic data evaluation system for the brand value of agricultural products, aiming to solve the technical problems in existing agricultural product brand evaluations, such as the lack of objective physical data support, difficulty in quantifying environmental impact, and inability to update dynamically in real time.

[0006] The system includes: a heterogeneous data acquisition and alignment module, a semantic-physical vector mapping engine, a dual-track entropy calculation module, an entropy difference value assessment module, and a dynamic update and storage terminal.

[0007] The heterogeneous data acquisition and alignment module is used to establish communication connections between the system and external data sources and perform spatiotemporal benchmark alignment operations. This module establishes a dual monitoring logic for the target brand domain and the environmental context domain: in the target brand domain, it collects publicly available marketing text data and real-time physical status data of the product to be evaluated; in the environmental context domain, it collects physical status data of non-branded ordinary agricultural products in the same spatiotemporal environment as the target brand. Based on a unified timestamp and geospatial coordinates, this module synchronously processes the aforementioned heterogeneous data, using linear interpolation or zero-order hold interpolation algorithms to unify sampling data of different frequencies onto a standard time axis, thereby providing a data foundation based on the same spatiotemporal cross-section for subsequent differentiated evaluation.

[0008] The semantic-physical vector mapping engine is connected to the heterogeneous data acquisition and alignment module to convert unstructured text into structured physical parameters. The engine has a pre-built semantic-physical mapping rule library, which includes explicit numerical mapping mechanisms and implicit semantic mapping mechanisms. The engine parses unstructured marketing text, identifies quality attribute entities and attribute modifier entities, and constructs semantic tuples. For qualitative descriptive terms, the engine invokes the implicit semantic mapping mechanism to query the corresponding physicochemical indicator thresholds, transforming the qualitative description into a commitment feature vector containing specific physical parameter threshold ranges or functional relationships, thereby establishing a physical benchmark for measuring the brand's fulfillment of its obligations.

[0009] The dual-track entropy calculation module is connected to the heterogeneous data acquisition and alignment module and the semantic-physical vector mapping engine, respectively, to perform dual-channel information entropy calculation to quantify the uncertainty of the system. This module first calculates the comprehensive deviation index by comparing the measured values ​​of the target product's physical state data with the constraints of the promised feature vector, generating a normalized deviation, and then performing a weighted summation based on a preset weight vector.

[0010] In the first calculation channel, this module calculates brand consistency entropy. Specifically, it acquires a comprehensive deviation index sequence within a set time window, fits the probability density function of this index using kernel density estimation, and then performs an integral operation on this probability density function. The value of brand consistency entropy reflects the degree of dispersion and volatility of brand delivery quality relative to its promises.

[0011] In the second calculation channel, this module calculates the environmental background entropy. Specifically, it obtains the mean and dispersion information of the physical state data of non-branded agricultural products to construct a background state vector; based on this vector, it performs Monte Carlo simulation to generate a set of virtual background physical state vectors, and then constructs the environmental background probability distribution and performs integral calculations. The numerical value of the environmental background entropy reflects the industry's average quality dispersion under the current spatiotemporal environment, constrained by natural conditions and logistical infrastructure.

[0012] The entropy difference value assessment module is connected to the dual-track entropy calculation module and is used to quantify the net gain of the target brand against environmental interference based on information theory principles. This module calculates the difference between the environmental background entropy and the brand consistency entropy as the original entropy difference component. To eliminate assessment bias caused by environmental factors, this module introduces an environmental pressure compensation factor to calculate the entropy difference gain. The calculation logic is as follows: when the environmental background entropy value is large, the environmental pressure compensation factor approaches one, thus retaining the original entropy difference component to characterize the target brand's ability to maintain quality stability in a high-entropy background environment; when the environmental background entropy value is small, the environmental pressure compensation factor approaches zero, thus suppressing the original entropy difference component and preventing calculation result bias caused by the stability of the environment itself. This module further performs a time-series-based cumulative evaluation, reads the cumulative reputation score of the previous evaluation period and uses an exponential decay model for time decay processing, superimposes the entropy difference gain obtained in this calculation onto the decayed historical score, and finally uses a logistic function to nonlinearly map the cumulative reputation score to a closed interval to generate a dynamic brand value index.

[0013] The dynamic update and storage terminal is connected to the entropy difference value assessment module to achieve smoothing and persistent storage of data. This terminal performs sliding window mid-range filtering preprocessing on the received brand value index sequence to remove instantaneous noise interference, and calculates the smoothed brand value index based on an exponentially weighted moving average. The terminal employs an event-driven mechanism, triggering a data write operation only when the difference between the smoothed brand value index and the previous stored value exceeds a preset threshold, or when the time since the last update exceeds the maximum silent period. This constructs a time-series storage structure containing historical data and outputs the assessment results through a standard data interface.

[0014] A second aspect of this invention provides a dynamic data evaluation method for the brand value of agricultural products. This method, based on the aforementioned system, includes the following steps:

[0015] Step S100: Establish dual data monitoring domains for the target and background, and perform spatiotemporal data alignment. By defining parallel data acquisition domains, textual and physical data of the target brand and physical data of the environmental background are acquired respectively, and structured target product physical state vectors and background state vectors are generated based on a unified spatiotemporal benchmark.

[0016] Step S200: Parse publicly available marketing text data, construct a marketing commitment feature vector containing physical parameter constraints, and perform semantic anchoring. Utilize natural language processing techniques to extract key quality attributes, and transform qualitative descriptions into quantitative allowable threshold ranges for physical parameters through semantic-physical mapping.

[0017] Step S300: Generate the physical delivery state vector and the background state vector. Based on the sensor measurement data and the statistical characteristics of the background samples, respectively, these vectors characterize the real-time physical attributes of the target product and the industry average physical state under the current environment.

[0018] Step S400: Calculate the dynamic deviation index. Compare the physical delivery status vector with the marketing commitment feature vector, calculate the normalized deviation of each attribute, and combine the attribute weights to generate a comprehensive deviation index reflecting the brand's fulfillment level.

[0019] Step S500: Perform dual-track entropy calculation. On one hand, construct a probability density function based on the comprehensive deviation index sequence and calculate the brand consistency entropy through integration. On the other hand, construct a background distribution model based on the background state vector and calculate the environmental background entropy through integration.

[0020] Step S600: Generate entropy difference gain and brand value index based on the entropy difference model. Calculate the difference between environmental background entropy and brand consistency entropy to quantify the brand's deterministic gain relative to the environmental benchmark. In this process, an environmental pressure compensation model is used to eliminate the computational gain caused by low-entropy background, and based on a time-series cumulative model, the current gain is integrated into the historical reputation score that has decayed over time. A dynamic brand value index for the current moment is generated through nonlinear mapping.

[0021] Step S700: Perform a time-series update of the brand value. A smoothing algorithm is used to denoise and extract trends from the generated dynamic brand value index, and the results are stored in the system database to form a continuous value evolution sequence.

[0022] This invention provides a dynamic data evaluation system and method for the brand value of agricultural products. It has the following beneficial effects:

[0023] 1. This invention uses a semantic-physical vector mapping engine to transform abstract marketing text promises into specific physical parameter threshold ranges and directly uses physical state data collected by sensors to calculate deviations. This method overcomes the limitations of traditional agricultural product brand evaluation, which relies too much on subjective questionnaires, financial statements, or expert scores. It lowers the evaluation dimension to the micro-level of physical delivery, ensuring that the brand value index directly reflects the actual performance quality and physical stability of agricultural products in the circulation process.

[0024] 2. This invention utilizes a dual-track entropy calculation and environmental pressure compensation mechanism to establish a synchronous monitoring and comparison logic between the target brand and the environmental background. By calculating the difference between the brand consistency entropy and the environmental background entropy, the system can automatically distinguish whether product quality fluctuations are caused by irresistible external environmental factors or by insufficient brand management capabilities. This mechanism avoids the brand being wrongly devalued due to adverse environments or overvalued due to superior environments, thereby accurately quantifying the brand's net effectiveness in combating environmental uncertainties.

[0025] 3. This invention adopts a time-series-based cumulative evaluation model and exponential decay mechanism, which changes the traditional static evaluation mode. The system assigns higher weight to recent performance data when calculating the current value through a continuously updated time-series value ledger, while retaining the cumulative effect of historical reputation. This dynamic update mechanism enables the evaluation results to keenly capture the real-time trend of brand quality changes, providing timely and traceable data support for supply chain supervision and consumer decision-making. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the structure of a dynamic data evaluation system for agricultural product brand value according to the present invention;

[0027] Figure 2 This is a schematic diagram illustrating the process of a dynamic data evaluation method for agricultural product brand value according to the present invention.

[0028] Among them, 101, heterogeneous data acquisition and alignment module; 102, semantic-physical vector mapping engine; 103, dual-track entropy calculation module; 104, entropy difference value assessment module; 105, dynamic update and storage terminal. Detailed Implementation

[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] See attached document Figure 1 This invention provides a dynamic data evaluation system for agricultural product brand value. The system is configured to perform real-time quantitative evaluation of agricultural product brand value based on a semantic anchoring and environmental background entropy difference model. The system includes: a heterogeneous data acquisition and alignment module 101, a semantic-physical vector mapping engine 102, a dual-track entropy calculation module 103, an entropy difference value evaluation module 104, and a dynamic update and storage terminal 105.

[0031] The heterogeneous data acquisition and alignment module 101 is configured to establish communication connections between the system and external data sources. These external data sources include publicly available marketing information databases of the target brand, IoT sensor networks across the entire supply chain, and environmental monitoring networks. The heterogeneous data acquisition and alignment module 101 performs spatiotemporal benchmark alignment operations, synchronizing the collected text data, target product physical state data, and non-branded agricultural product physical state data based on a unified timestamp and geospatial coordinates.

[0032] The semantic-physical vector mapping engine 102, connected to the heterogeneous data acquisition and alignment module 101, is configured to receive pre-processed marketing text data. The semantic-physical vector mapping engine 102 has a pre-built natural language processing unit and a physical parameter mapping rule base for parsing unstructured marketing text and extracting key quality attributes. The semantic-physical vector mapping engine 102 converts the key quality attributes into commitment feature vectors containing specific physical parameter threshold ranges or functional relationships.

[0033] The dual-track entropy calculation module 103 is connected to the heterogeneous data acquisition and alignment module 101 and the semantic-physical vector mapping engine 102, respectively. The dual-track entropy calculation module 103 is configured to receive the commitment feature vector, the real-time physical delivery status vector of the target product, and the environmental background status vector.

[0034] The dual-track entropy calculation module 103 contains two independent calculation channels. The first channel is used to calculate brand consistency entropy by comparing the promised feature vector with the physical delivery state vector, generating a deviation probability distribution, and calculating the corresponding information entropy value. The second channel is used to calculate environmental background entropy by calculating the information entropy value reflecting the current level of natural environmental degradation by statistically analyzing the dispersion of physical state data of non-branded agricultural products in the environmental background monitoring network.

[0035] The entropy difference value assessment module 104 is connected to the dual-track entropy calculation module 103. The entropy difference value assessment module 104 is configured to receive brand consistency entropy and environmental background entropy, and execute the difference operation logic. Based on the difference between the environmental background entropy and the brand consistency entropy, the entropy difference value assessment module 104 calculates the entropy difference gain at the current moment, and combines it with historical accumulated reputation points to generate a dynamic brand value index.

[0036] The dynamic update and storage terminal 105 is connected to the entropy difference value assessment module 104. The dynamic update and storage terminal 105 is configured to receive the dynamic brand value index according to a preset time step, append the dynamic brand value index to the historical assessment data to update the brand value index sequence, construct an immutable time series value ledger, and output the current brand value assessment result to the outside world through a standard data interface.

[0037] In the system's data flow logic, the heterogeneous data acquisition and alignment module 101 transmits the acquired unstructured data to the semantic-physical vector mapping engine 102, while simultaneously transmitting the structured physical sensor data to the dual-track entropy calculation module 103. The standardized vector parameters processed by the semantic-physical vector mapping engine 102 are then input as baseline data to the dual-track entropy calculation module 103. After completing the bidirectional entropy calculation, the dual-track entropy calculation module 103 transmits the numerical results to the entropy difference value assessment module 104 for final value synthesis.

[0038] See attached document Figure 2 This invention provides a dynamic data evaluation method for the brand value of agricultural products. This method is executed collaboratively by a heterogeneous data acquisition and alignment module 101, a semantic-physical vector mapping engine 102, a dual-track entropy calculation module 103, and an entropy difference value evaluation module 104. Specifically, it includes the following steps:

[0039] Step S100: Establish a dual data monitoring domain for both the target and background information, and perform spatiotemporal data alignment. The system defines two parallel data acquisition domains: the target brand domain and the environmental background domain. Within the target brand domain, collect a set of publicly available marketing text data for the agricultural product to be evaluated. Public marketing text data set This includes product advertising, packaging labeling, and certification information; it also collects real-time physical status data of the brand's products along the logistics chain. In the background environmental domain, it collects physical status data of non-branded ordinary agricultural products in the same spatiotemporal environment as the target brand. The system synchronously processes the aforementioned text data, target product physical status data, and background physical status data based on a unified timestamp and geospatial coordinates, ensuring that subsequent calculations are based on the same physical process cross-section. Sensor calibration and network transmission protocols during the data acquisition process can be implemented using existing IoT communication standards by those skilled in the art, and will not be elaborated upon here.

[0040] Step S200: Construct a marketing commitment feature vector and perform semantic anchoring. The system utilizes natural language processing algorithms to process the publicly available marketing text dataset. Perform analysis and extraction The key quality attributes are denoted as the feature set. The system has a pre-built semantic-physical mapping rule base containing the correspondence between keywords and physical values. Each extracted feature is then processed through table lookup matching or regular expression extraction. This is transformed into an allowable threshold range or distribution function of physical parameters, thereby generating a marketing commitment feature vector. The vector is represented as:

[0041] ;

[0042] in, Indicates the first The constraints of the key quality attributes are expressed as intervals. ,in As the lower limit, This is the upper limit. For example, for descriptive terms, the system will map them to specific temperature value ranges.

[0043] Step S300: Generate the physical delivery state vector and the background state vector. Based on real-time monitoring data from the IoT sensor array, the system at time... Generate the physical delivery state vector of the target product :

[0044] ;

[0045] in For the first The system calculates the measured physical values ​​of several key quality attributes. Simultaneously, it calculates the statistical characteristics of samples in the background domain to generate a background state vector. This vector contains the mean and dispersion information of each attribute of the background sample, and is used to characterize the industry's average physical state under the current environment.

[0046] Step S400: Calculate the dynamic deviation index. The system compares the physical delivery status vector with the marketing commitment feature vector to quantify the target brand's fulfillment deviation across various attributes. Define the first... The measured physical values ​​of the key mass attributes at time 10:00 Normalization bias The calculation logic is as follows:

[0047] ;

[0048] Further combining the preset weight vectors of each attribute Calculate the comprehensive deviation index :

[0049] ;

[0050] The calculation of the comprehensive deviation index This reflects the extent to which the brand's actual delivery deviates from its promises.

[0051] Step S500: Perform dual-track entropy calculation. This step includes calculating brand consistency entropy. With environmental background entropy .

[0052] The system operates within a set time window. Internally, a comprehensive deviation index is constructed using kernel density estimation or histogram statistical methods. probability density function And calculate brand consistency entropy based on Shannon entropy definition. :

[0053] ;

[0054] Brand Consistency Entropy The lower the value, the more stable and consistent the brand's delivery quality is with its promises. Simultaneously, the system selects key indicators from background domain samples to construct a probability distribution. Calculate the environmental background entropy :

[0055] ;

[0056] Environmental background entropy Used to quantify the degree of disorder caused by the current external environment leading to the loss of natural quality.

[0057] Step S600 generates entropy difference gain and brand value index based on the entropy difference model. The system first calculates the difference between the environmental background entropy and the brand consistency entropy, quantifying the entropy difference gain of the target brand relative to the environmental background. Step S600 employs a nonlinear compensation model to eliminate artificially high components caused by easily controllable environments (low background entropy) when calculating the gain. Subsequently, the system, based on a time-series cumulative model, integrates the current entropy difference gain into the historical reputation score, generating time-series values ​​through nonlinear mapping. Brand dynamic value index This process ensures that the value assessment reflects both the quality of the current batch of deliveries and the historical continuity of the brand's reputation.

[0058] Step S700: Time-series update of brand value. The system iteratively executes the above steps according to discrete time steps and updates the brand value series using a smoothing algorithm. A smoothing factor is set. The value of the next moment The update rules are as follows:

[0059] ;

[0060] To ensure that the evaluation results reflect real-time changes while maintaining trend continuity, the final brand value sequence is stored in the system database for subsequent querying or retrieval.

[0061] In this embodiment, the specific execution process of the feature vector construction and mapping method based on semantic anchoring is as follows:

[0062] Step S210, targeting the publicly available marketing text data set Preprocessing and entity extraction are performed. The system receives raw text data from the heterogeneous data acquisition and alignment module 101, performs data cleaning to remove stop words and non-information characters. Subsequently, named entity recognition technology based on an agricultural ontology library is used to scan the cleaned text. The system identifies and extracts two types of key entities from the text:

[0063] One type is quality attribute entities, such as sugar content, firmness, fruit diameter, and cold chain temperature; the other type is attribute modification entities, such as >12, crisp and sweet, extra grade, and 0-4 degrees.

[0064] Step S211: Establish dependency pairs between attributes and modifiers. To accurately parse the commitment value corresponding to a specific attribute, the system uses a dependency parsing algorithm to process the extracted entities and construct the logical association between attributes and modifiers. The system generates several original semantic tuples. ,in Representing the One identified quality attribute, This represents the corresponding descriptive phrase. For example, from the text "Shaanxi Luochuan apples, crisp texture, sugar content above 13," we can extract... Texture: crisp and Sugar content, above 13 .

[0065] Step S212: Perform quantitative mapping based on semantic anchoring. This step aims to transform unstructured semantic tuples into computer-computable ranges of physical parameters. The system invokes a pre-built semantic-physical mapping rule base. This rule base contains two types of mapping mechanisms:

[0066] The first type is explicit numerical mapping. When When there are explicit numerical values ​​and comparison operators, the system directly extracts the numerical boundaries.

[0067] The second type is implicit semantic mapping. When When the term is a qualitative adjective, the system queries the rule base for the corresponding physicochemical index threshold. The physicochemical index thresholds in the rule base are pre-calibrated based on the statistical distribution intervals of national agricultural standards or historical physical testing data. The mapping relationship is defined as a function. For example, regarding the attribute "texture," if the modifier is "crisp," the function... Return to hardness index range [7.0 kg / cm] 2 9.5 kg / cm 2 This process transforms qualitative descriptive terms into quantitative physical parameter ranges.

[0068] Step S213: Eliminate conflicts and generate standardized constraints. In actual marketing texts, if there are multiple descriptions of the same attribute, the system executes conflict resolution logic. The priority rule is set as follows: explicit numerical descriptions have higher priority than implicit semantic descriptions. If conflicts exist, the intersection or the higher priority value is taken. After processing, for the system's predefined... Each of the key quality attributes , determine its final first Constraints of key quality attributes .

[0069] This constraint Formalized as a closed interval or a half-open interval:

[0070] ;

[0071] in, As the lower limit, This is the upper limit. If an attribute is not mentioned in the text, the system assigns it an industry-standard default value range, or marks it as an infinite range. .

[0072] Step S214: Assemble the marketing commitment feature vector. The system assigns constraints corresponding to all key quality attributes. Arranged according to a preset dimensional order, the final marketing commitment feature vector is generated. :

[0073] ;

[0074] The marketing commitment feature vector As the baseline input for the subsequent dual-track entropy calculation module 103, it represents the target parameter range of the brand in the physical dimension. The system will generate a marketing commitment feature vector. Stored in a temporary cache and tagged with a corresponding timestamp, to await matching with the real-time collected physical delivery status vector.

[0075] In this embodiment, the system acquires the physical basis data for subsequent entropy calculation by constructing a dual-channel monitoring mechanism. The specific execution steps are as follows:

[0076] Step S310: Deploy the target monitoring sensor array and collect real-time data. For the target brand agricultural products to be evaluated, the system configures IoT sensing devices at key nodes in the supply chain, including post-harvest pre-cooling warehouses, cold chain transport vehicles, and terminal shelves. The IoT sensing devices include temperature and humidity sensors and gas sensors for monitoring environmental parameters, and a triaxial accelerometer for monitoring mechanical damage risk. In addition, for appearance indicators such as fruit diameter and color, the system connects to the machine vision inspection interface or portable spectrometer data interface on the sorting line. The system collects raw data sequences according to a preset sampling frequency and uses a Kalman filter algorithm to remove instantaneous outliers caused by sensor noise. The preprocessed data is mapped to a predefined system mapping algorithm. Regarding key quality attributes, at time Generate physical delivery state vector :

[0077] ;

[0078] in, For the first The measured physical values ​​of several key quality attributes. For the specific selection and installation methods of the aforementioned sensors, those skilled in the art can refer to existing cold chain logistics monitoring technology standards, which will not be elaborated upon here.

[0079] Step S311: Define the background domain data source and perform control group sampling. To calculate the environmental background entropy, the system needs to acquire physical state data of non-branded ordinary agricultural products under the same spatiotemporal environment. The system acquires background data through two paths: Path 1 is physical control group monitoring, which involves placing ordinary bulk samples that have not undergone brand grading in the same transport vehicle or storage area as the target brand product, and deploying isomorphic sensing devices for synchronous monitoring; Path 2 is regional public data access, which involves accessing the open API of the wholesale market or public logistics platform in the production area to obtain the average physicochemical index data of similar agricultural products in the current time window. Based on the spatiotemporal alignment principle, the system filters out the set of background samples that are within the same geographical coordinate radius (e.g., within 5 kilometers) or the same logistics batch ID as the target product, denoted as... .

[0080] Step S312: Extract background statistical features and construct a background state vector. The system processes the background sample set. Statistical analysis was performed on the data to quantify the natural impact of the current environment on the quality of agricultural products. For each key quality attribute... The system calculates the mean of the background sample set. with standard deviation These statistics constitute the background state vector. :

[0081] ;

[0082] The background state vector It does not directly represent a single physical value, but rather includes a set of parameters needed to construct the probability distribution of the environmental background. In subsequent entropy calculation steps, the system... The mean and standard deviation are used to reconstruct the physical state distribution model of the background domain using the Gaussian distribution assumption or kernel density estimation, thereby providing a basis for calculating the environmental background entropy. Provide data support.

[0083] Step S313: Time synchronization and structured storage of data. Since the target data and background data may come from different sources, their sampling frequencies may differ. The system performs linear interpolation or zero-order hold interpolation to unify the two sets of data onto a standard time axis. The system will then... and The system associates data with the database and creates an index to ensure that when calculating the entropy difference at a certain moment, the brand data and background data are strictly in the same spatiotemporal environment.

[0084] In this embodiment, the dual-track entropy calculation module 103 maps real-time monitoring data from the physical world into quantitative indicators reflecting the degree of fulfillment of brand commitments by performing the following steps:

[0085] Step S410: Receive and parse heterogeneous vector data. The system first obtains the marketing commitment feature vector from the semantic-physical vector mapping engine 102. And obtain the physical delivery state vector from the heterogeneous data acquisition and alignment module 101. The system performs dimension verification on the two vectors to ensure that they contain the correct dimensions. The key quality attributes are completely consistent in order and definition. If there is a missing dimension (e.g., sensor data packet loss), the system calls the preset null value filling strategy, using the valid value of the previous time step to fill in the data or a linear interpolation algorithm based on adjacent time steps, so as not to produce bias contribution, or to mark the data at that time step as invalid.

[0086] Step S411: Perform normalized deviation calculation for each single attribute. To eliminate differences between different physical dimensions, the system performs normalized deviation calculation for each key mass attribute. Calculate the dimensionless normalized deviation. System extraction. The corresponding number in Constraints of key quality attributes ,as well as The corresponding number in Measured physical values ​​of key quality attributes .

[0087] The system calculates the first step based on the following piecewise function. The measured physical values ​​of the key mass attributes at time 10:00 Normalization bias :

[0088] ;

[0089] in, As the lower limit, The upper limit, For a preset numerical stability constant (e.g.) This calculation logic ensures that the deviation is zero when the measured value falls within the promised range; when the measured value exceeds the upper limit or falls below the lower limit, the deviation increases linearly. A constant is introduced. This is to prevent calculation errors caused by dividing by zero when the physical boundary value is 0.

[0090] Step S412: Determine attribute weights based on semantic saliency. To accurately reflect the impact of brand marketing priorities on value assessment, the system does not use average weights, but instead relies on publicly available marketing text data sets. The word frequency statistics results dynamically generate preset weight vectors. The system uses inverted indexes or TF term frequency statistics algorithms to statistically analyze each key quality attribute. The frequency of their synonyms in the text collection And calculate its proportion in the total frequency of all attributes. Preset weight vector Middle The formula for calculating each element is:

[0091] ;

[0092] This ensures that quality attributes that appear frequently in marketing texts are given higher weight in the final value assessment. If an attribute does not appear explicitly in the text but is monitored as a basic indicator (such as pesticide residue detection), the system assigns it a preset minimum weight (e.g., 0.01).

[0093] Step S413: Synthesize the comprehensive deviation index. The system performs a weighted summation of the normalized deviation of each attribute and its corresponding weight to generate the time index. Comprehensive Deviation Index :

[0094] ;

[0095] The comprehensive deviation index It is a non-negative scalar, and its physical meaning is: at the current moment The degree to which the overall physical condition of the target product deviates from its brand marketing promise. The system will generate... The sequence is stored in a time-series database as the basic input data for subsequently constructing a probability density function to calculate brand consistency entropy.

[0096] In this embodiment, the dual-track entropy calculation module 103 calculates the entropy value reflecting the brand's performance stability by transforming the discrete comprehensive deviation index sequence into a continuous probability distribution. The specific steps are as follows:

[0097] Step S510: Construct the deviation sample set within the time window. The system sets a statistical time window. This window can be a time-based sliding window (e.g., the past 24 hours) or a batch-based fixed window. The system extracts data from the time-series database within this time window. All internally generated comprehensive deviation indices Data points to construct a deviation sample set ,in This represents the total number of sampling points within the window. To reduce computational noise, the system performs outlier detection on the sample set using the Laida criterion (3). Criteria) Remove statistical outliers (i.e. Data points, among which and These are the mean and standard deviation of the sample set, respectively.

[0098] Step S511: Construct a probability density function based on kernel density estimation. Since the sampled deviation data is discrete, the system uses kernel density estimation (KDE) to fit the probability density function of the comprehensive deviation index in order to calculate continuous entropy. Compared to simple histogram statistics, KDE provides a smoother and more accurate description of the distribution. Its calculation formula is as follows:

[0099] ;

[0100] in, As a deviation variable, The bandwidth parameter determines the smoothness of the distribution curve; For the kernel function, in this embodiment, the Gaussian kernel function is selected:

[0101] ;

[0102] For bandwidth The system automatically calculates the selection based on Silverman's rule of thumb:

[0103] ;

[0104] in For the biased sample set The sample standard deviation. This transforms discrete deviation records into a continuous function describing the probability distribution of the deviation. .

[0105] Step S512: Perform numerical integration calculation of the entropy value. Based on Shannon's definition of information entropy, the system calculates brand consistency entropy using the complex trapezoidal rule or Simpson's rule. The calculation formula is:

[0106] ;

[0107] At the computer execution level, the system will integrate the intervals. Discretize into small step sizes , The value is the bandwidth parameter. From 1 / 100 to 1 / 10, convert the above integrals into summation operations:

[0108] ;

[0109] in, The value is set to 1.2 times the maximum value of the biased sample set, and is used to cover the tail portion of the distribution.

[0110] Step S513, mapping the physical meaning of entropy. The calculated brand consistency entropy. It quantifies the uncertainty of physical quality during the brand's delivery process. If Approaching 0 indicates a probability distribution A high concentration of values ​​around a single threshold indicates a brand's exceptionally high consistency and predictability in fulfilling its promises; if A larger value indicates a flat or multi-peaked deviation distribution, suggesting high dispersion in the system's internal physical state and unstable quality control capabilities. The calculation result is transmitted in real-time to the entropy difference value assessment module 104.

[0111] In this embodiment, the system constructs a virtual comparison model of the background environment and calculates the natural entropy value under no-brand intervention conditions, using this as a baseline noise level for evaluating brand management effectiveness. The specific steps are as follows:

[0112] Step S520: Obtain background state statistical features and constraint parameters. The system retrieves the current time window from the database. Background state vector within This vector contains background samples. Mean of key quality attributes with standard deviation At the same time, the system reads the marketing commitment feature vector. and preset weight vector The purpose of this step is to ensure that the assessment criteria for the environmental background and the assessment criteria for the target brand use the same characteristic dimensions and weighting system, thereby guaranteeing the comparability of the assessment results.

[0113] Step S521: Generate a background deviation sample set based on Monte Carlo simulation. Since the background state exists in the form of statistical parameters, the system performs Monte Carlo simulation to obtain a deviation distribution isomorphic to the brand side. The system assumes that the physical properties in the background environment follow independent Gaussian distributions, based on... The mean and standard deviation recorded in the data are generated using the Box-Muller transform or the Ziggurat algorithm. One (e.g.) Virtual background physical state vector. For each virtual vector, the system repeatedly executes the deviation calculation logic to calculate the feature vector relative to the marketing commitment. The weighted deviation value. Therefore, the system generates a value containing... Background deviation sample set of data points Among them, variables It represents the degree of deviation of ordinary agricultural products in the context of the target brand promise.

[0114] Step S522: Construct the environmental background probability distribution. The system uses kernel density estimation to process the background deviation sample set. Construct the probability distribution function of the background domain The system uses the same kernel function type as in the brand consistency entropy calculation, but independently calculates the bandwidth parameter suitable for background data. This distribution This reflects the natural distribution pattern of product quality deviating from the ideal promised value under current logistical conditions and natural environment, without the application of specific brand management controls.

[0115] Step S523: Calculate the environmental background entropy. The system performs an integral operation on the constructed probability distribution to quantify the degree of disorder in the environmental background. The system calculates the environmental background entropy. :

[0116] ;

[0117] Although deviation variables While physically non-negative, to maintain the generality of the mathematical model and consistency with the formulas defined in the aforementioned embodiments, the lower limit of integration is formally written as... In the specific implementation of numerical calculation, the system sets the integration interval to... ,in For 1.2 times the maximum value of the background deviation sample set, The region, probability density Set to 0.

[0118] Environmental background entropy It is used as a denominator or subtrahend in subsequent value assessments. Its physical meaning lies in representing the environmental disorder benchmark, that is, the unavoidable dispersion of agricultural products caused by the current spatiotemporal environment. If... A higher score indicates poor external environmental conditions (e.g., high temperatures, bumpy roads leading to significant natural quality dispersion). If the brand can maintain a low brand consistency entropy under these conditions, it means the brand's quality control system has achieved significant effectiveness and should receive a higher value score; conversely, if... The lower value indicates that the environment itself is conducive to maintaining quality. The brand's low entropy performance is the norm in the industry, and its relative value contribution is small.

[0119] In this embodiment, the system quantifies the net contribution of brand intervention to reducing system uncertainty based on the mutual information principle in information theory. The specific execution steps are as follows:

[0120] Step S610: Temporal alignment and validity verification of entropy data. The system receives the current time window from the dual-track entropy calculation module 103. Brand consistency entropy within With environmental background entropy The system performs data integrity checks to ensure that the two entropy values ​​are calculated based on the same batch of physical data and the same spatiotemporal background data. If If the value is zero or missing (e.g., in a highly controlled laboratory environment with no natural background), the system calculates the theoretical maximum entropy value based on a uniform distribution. Substitutes are made, and the calculation formula is as follows: ,in This represents the total number of discretized intervals, or it pauses the current calculation and marks it as lacking a background reference.

[0121] Step S611: Calculate the original entropy difference components. The system performs core difference operations to determine the degree of system disorder eliminated by brand management actions. Original entropy difference components The calculation formula is as follows:

[0122] ;

[0123] in, This represents a natural, discrete state without brand intervention. This represents the actual delivery status under the brand's commitment. Physically, this represents information gain, specifically the improvement in quality consistency between branded and non-branded products. If... This indicates that the brand has effectively reduced uncertainty; if This indicates that brand management has failed, and the quality stability is even lower than the natural level of the environment.

[0124] Step S612: Determine the category value sensitivity coefficient. Different categories of agricultural products have different sensitivities to consistency value. The system queries the pre-set category sensitivity database and extracts the corresponding category value sensitivity coefficient based on the SKU category ID of the currently evaluated product. Value sensitivity coefficient for this category It is a dimensionless positive real number, and its value is based on the reciprocal of the Pearson correlation coefficient between the volatility of the average market transaction price of this product category and the variance of the product's physical indicators within a historical time window.

[0125] Step S613: Calculate the entropy difference gain exponent. The system combines the original entropy difference components with the category value sensitivity coefficient to construct an entropy difference gain model and calculate the final entropy difference gain. Considering that negative entropy differences should not generate positive value, and that the impact of environmental benchmarks on gain difficulty needs to be taken into account, the system adopts the following nonlinear correction formula:

[0126] ;

[0127] in: For linear rectifier functions, defined as follows: This function ensures that when When the gain is zero, it avoids generating incorrect value incentives. This is an environmental stress compensation factor. For example, the preset adjustment constant (e.g.) The role of this environmental pressure compensation factor is: when the environmental background entropy... When the value is large (high environmental dispersion, making control difficult), Approaching 0, the entire compensation factor When the value approaches 1, the system retains the full amount of the original entropy difference component to acknowledge the brand's efforts to maintain quality in harsh environments; when When the numerical value is extremely small (the environment itself is stable, and the natural dispersion is low), As the value approaches 1, the entire compensation factor approaches 0, thereby suppressing the artificially high scores caused by the ease of controlling the environment itself. The final output is a scalar value that represents the deterministic value measure created by the brand in the current batch of deliveries through quality control measures, after noise reduction and weight correction.

[0128] Step S614, persistent storage of the gain exponent. The system will calculate and store the gain exponent. Associated with the batch ID, this value is stored in a local relational database or distributed ledger. This value will serve as the direct basis for the pricing system to trigger dynamic pricing or credit score settlement.

[0129] In this embodiment, the system does not rely on the evaluation results at a single moment, but instead establishes a time-series-based cumulative evaluation model. By simulating the decay and cumulative effects of market memory, it generates a dynamically evolving brand value index. The specific execution steps are as follows:

[0130] Step S620: Initialization and Historical Status Loading. Based on the unique identifier of the brand to be evaluated, the system reads the accumulated reputation score from the time-series database for the previous evaluation period. If this is the brand's first time connecting to the system, the system will... Initialize to the preset base reputation value (For example, set to 0). At this time, the system simultaneously reads the entropy differential gain obtained from this calculation. This value is calculated using the entropy difference gain model in the aforementioned embodiments.

[0131] Step S621: Perform time decay calculation. Considering that market memory of brand quality is time-sensitive and brand value requires continuous quality delivery to maintain, the system introduces a memory retention coefficient. Historical reputation is attenuated. Memory retention coefficient. The values ​​are determined based on the exponential decay model: ,in The industry forgetting rate constant, This is the time interval between two assessments. For high-frequency, fast-moving consumer goods, a larger interval is set. The value decays at a faster rate, requiring brands to maintain high-frequency and stable output. The historical integral after decay is...

[0132] Step S622: Update the accumulated reputation score. The system will use the entropy difference gain obtained in this evaluation period. Superimpose the superposition onto the decayed historical integral to calculate the current time step. Accumulated credit points The calculation formula is as follows:

[0133] ;

[0134] The formula reflects the dynamic balance mechanism of brand value building: if the entropy difference gain is consistently 0 or negative (unstable quality or no output), the accumulated reputation score will decay exponentially over time; only a continuous and stable positive gain input can offset the time decay term and drive the score growth.

[0135] Step S623: Nonlinear mapping generates the brand value index. This is due to accumulated reputation points. Theoretically there is no upper limit. However, for easier intuitive display and cross-category horizontal comparison, the system uses a logistic function to... Mapping to closed intervals Generate the final brand value index. The calculation formula is:

[0136] ;

[0137] in, Maximum exponent limit (e.g., 100); The curve steepness coefficient (e.g., 0.1) controls the sensitivity of the exponent to changes in the integral. The inflection point threshold represents the point reached. The integral required to achieve the score. This mapping logic introduces a sensitivity saturation characteristic, meaning that as the score increases, the increment of integral required to improve the score increases non-linearly, thus mathematically simulating the increasing difficulty from "good" to "excellent" in brand building.

[0138] Step S624, Index Update and Trend Analysis. The system will calculate the generated brand value index. The value ledger is written into the brand account. The system also calculates the short-term trend slope of the index, specifically using the least squares method on past... a point in time (e.g.) )of The system performs linear fitting on the sequence and extracts the slope parameter of the fitted line. If the slope is significantly negative and the duration exceeds a preset threshold, the system generates an early warning signal, indicating to the brand manager that the current physical delivery quality can no longer support the brand reputation and that there is a risk of value depreciation.

[0139] In this embodiment, to avoid unexpected fluctuations in the evaluation results due to accidental fluctuations in a single monitoring data, which could affect the stability of the downstream pricing system or reputation display, the system introduces a multi-level smoothing process and an event-driven update mechanism. The specific execution steps are as follows:

[0140] Step S710: Perform sliding window mid-range filtering preprocessing. The system receives the brand value index in real time. To address potential impulse noise or outliers caused by packet loss and retransmission in sensor networks, the system first performs noise reduction. The system constructs a data structure of length... (For example A sliding time window is used to extract the historical index sequence within that window. The system calculates the median of the sequence and generates a preprocessing value index. The specific sorting algorithm for median calculation can be implemented using quicksort or insertion sort, which are well-known techniques in the field and will not be elaborated upon here. This step effectively eliminates statistical outliers while preserving the step characteristics of the signal, ensuring the system remains responsive to real value mutations.

[0141] Step S711: Trend smoothing based on exponentially weighted moving average. To obtain a stable indicator that reflects the long-term value trend of the brand, the system performs preprocessing of the value index. Perform an index-weighted moving average calculation to smooth the brand value index. The calculation formula is as follows:

[0142] ;

[0143] in, This is a smoothing coefficient, with a value ranging from (0,1). This coefficient determines the system's sensitivity to new data. In this embodiment, It is not a fixed constant, but rather based on the entropy difference gain. The absolute value is adaptively adjusted using a piecewise linear mapping function. The specific calculation logic is as follows: A base smoothing coefficient is set. (e.g., 0.05) and gain sensitivity coefficient (e.g., 0.1), calculate If the calculation result exceeds the preset upper limit (e.g., 0.3), then let This mechanism ensures that when a significant quality event occurs (i.e., When the system is larger, the system increases. Values ​​are updated rapidly to reflect the value state; when in a steady state (i.e., ... When the size is small, the system decreases. The value is used to suppress random disturbances.

[0144] Step S712: Execute event-driven dynamic update determination. To reduce the write pressure on the blockchain notarization module or high-concurrency database, the system does not persist the calculation results at every moment. Instead, it uses an event-driven mechanism to determine whether to trigger an update operation. The system sets two triggering conditions:

[0145] Condition 1: Value change threshold triggered. The system calculates the smoothed brand value index at the current moment. The value index of records stored in the last persistent storage The absolute value of the difference between them. If ,in If the minimum change threshold is set (e.g., 0.5 points), an update will be triggered.

[0146] Condition 2: Forced refresh time is triggered. The system records the timestamp of the last update. If the current time satisfy ,in If the maximum silent period is set (e.g., 24 hours), an update will be forcibly triggered.

[0147] Step S713, Output and Persistence. When any of the above triggering conditions are met, the system will output and persist the current data. The latest valid brand value is written into the corresponding database field, and a state change notification is sent to the smart contract or front-end component that subscribes to this data. This mechanism ensures that the system optimizes data storage and network transmission resource usage while maintaining the real-time validity of the data.

[0148] In this embodiment, the system is configured to evaluate the brand fulfillment capability and dynamic value of a target brand during cross-regional cold chain transportation. The specific execution steps are as follows:

[0149] Initialize marketing commitments and monitoring dimensions. The system first reads the target brand's electronic contracts or product specifications and extracts the marketing commitment feature vector. In this scenario, key quality attributes Set as 3 dimensions: Represents the refrigeration temperature (promised range) ), Represents relative humidity (commitment range) ), Represents the amplitude of transport vibration (promised upper limit) The system also sets a preset weight vector. Based on the degree of influence of fruit preservation characteristics on different dimensions, the temperature dimension is assigned the highest weight (e.g., ).

[0150] Collect physical delivery data and environmental background data. During transportation, smart sensor terminals deployed inside the cargo container collect real-time physical data at a preset frequency (e.g., once every 10 minutes) to form a physical delivery status vector. Simultaneously, the system obtains the vehicle's current location via onboard GPS positioning and combines this with a meteorological service interface to acquire data on external ambient temperature and road condition levels. Using this location and environmental data as an index, the system retrieves a pre-set environmental benchmark database, matches and extracts the statistical characteristic parameters (mean and standard deviation) corresponding to ordinary logistics under those environmental conditions, thereby constructing a background state vector. For example, when querying the ambient temperature at which the vehicle was driving. And when the road surface is bumpy and unpaved, the extracted This includes a large mean temperature and standard deviation of vibration, reflecting the high physical state dispersion of the background environment.

[0151] Perform dual-track entropy calculation: On one hand, the system calculates brand consistency entropy. If the target brand uses modified atmosphere packaging and suspended shock-absorbing refrigerated trucks, the measured temperature remains consistently stable at [temperature value missing]. Vibration amplitude controlled within The following is the calculated result. The value is low (e.g.) This characterizes the high concentration of physical delivery states in the probability distribution; on the other hand, the system's computational environment background entropy... The system is based on The parameters in the simulation are used to generate a virtual cargo state distribution without special protective measures. In this context, the distribution curve of the virtual data is flat and has a long tail, resulting in... A high numerical value (e.g., 2.8) indicates a high degree of dispersion of the baseline state under the current transportation conditions.

[0152] Calculate the entropy difference gain and update the value index. The system calculates the entropy difference gain. Due to environmental background entropy (2.8) Significantly greater than brand consistency entropy (0.15), and the category value sensitivity coefficient of the target fresh produce category Setting the value to a high level, the system calculates a significant positive gain. Physically, this indicates that, under harsh environmental conditions, the target brand, through the effective intervention of its quality control system, achieves significantly better product condition certainty than the environmental baseline. Subsequently, the system updates the brand value index using this positive gain based on an evolutionary model. , making Numerical improvement.

[0153] A dynamic settlement application based on a value index. The system will update the brand value index. Input is sent to the intelligent settlement module. This module executes the tiered settlement coefficient matching logic: if... In the range Set settlement coefficient The final settlement unit price is the contract benchmark price multiplied by ;like In the range ,set up ;like Below 60, set This triggers a default determination. Through the above mechanism, the system directly maps the stability data of physical delivery quality to transaction execution parameters. The specific hardware implementation of the above data acquisition, transmission, and calculation can be achieved by those skilled in the art using existing microcontrollers and IoT communication modules, which are well-known technologies in the field and will not be elaborated upon here.

Claims

1. A dynamic data evaluation system for the brand value of agricultural products, characterized in that, It includes a heterogeneous data acquisition and alignment module (101), a semantic-physical vector mapping engine (102), a dual-track entropy calculation module (103), an entropy difference value assessment module (104), and a dynamic update and storage terminal (105). The heterogeneous data acquisition and alignment module (101) is configured to establish a communication connection between the system and an external data source, perform a spatiotemporal reference alignment operation, and synchronize the acquired text data, target product physical state data and non-branded background agricultural product physical state data according to a unified timestamp and geospatial coordinates. The semantic physical vector mapping engine (102) is connected to the heterogeneous data acquisition and alignment module (101) and is configured to parse unstructured marketing text to extract key quality attributes and convert the key quality attributes into commitment feature vectors containing specific physical parameter threshold ranges or functional relationships. The dual-track entropy calculation module (103) is connected to the heterogeneous data acquisition and alignment module (101) and the semantic physical vector mapping engine (102) respectively. It is configured to calculate the brand consistency entropy through the first calculation channel and the environmental background entropy through the second calculation channel based on the commitment feature vector, the target product physical state data and the non-brand background agricultural product physical state data. The entropy difference value assessment module (104) is connected to the dual-track entropy calculation module (103) and is configured to calculate the entropy difference gain at the current moment based on the difference between the environmental background entropy and the brand consistency entropy, and generate a dynamic brand value index by combining the historical accumulated reputation score. The dynamic update and storage terminal (105) is connected to the entropy difference value assessment module (104) and is configured to receive the dynamic brand value index, add the dynamic brand value index to the historical assessment data to update the brand value index sequence, construct a time series value ledger, and output the assessment results to the outside world through a standard data interface.

2. The dynamic data evaluation system for agricultural product brand value according to claim 1, characterized in that, The monitoring domain defined by the heterogeneous data acquisition and alignment module (101) includes the target brand domain and the environmental background domain; Collect the text data and the physical status data of the target product in the target brand domain; In the aforementioned environmental background domain, physical state data of non-branded background agricultural products that are in the same spatiotemporal environment as the target brand are collected as background data; The heterogeneous data acquisition and alignment module (101) uses linear interpolation or zero-order hold interpolation to unify the text data, the target product physical state data and the background data onto a standard time axis.

3. The dynamic data evaluation system for agricultural product brand value according to claim 1, characterized in that, The semantic-physical vector mapping engine (102) has a pre-built semantic-physical mapping rule library, which includes an explicit numerical mapping mechanism and an implicit semantic mapping mechanism. The semantic-physical vector mapping engine (102) identifies the key quality attributes in the unstructured marketing text, which are divided into quality attribute entities and attribute modification entities; When the attribute modifies an entity as a qualitative descriptor, the implicit semantic mapping mechanism is used to query the corresponding physicochemical index threshold in the semantic-physical mapping rule base, transforming the qualitative description into an allowable threshold range of physical parameters, thereby generating the commitment feature vector.

4. The dynamic data evaluation system for agricultural product brand value according to claim 1, characterized in that, Before calculating the brand consistency entropy, the dual-track entropy calculation module (103) is configured to first calculate the comprehensive deviation index. The calculation steps of the comprehensive deviation index are as follows: The dual-track entropy calculation module (103) compares the measured physical values ​​in the target product physical state data with the constraints in the promised feature vector; For each of the key quality attributes, when the measured physical value falls within the range of the constraints, the normalized deviation is recorded as zero. When the measured physical value exceeds the upper limit of the constraint interval or falls below the lower limit of the constraint interval, the normalization deviation increases linearly with the degree of deviation. The dual-track entropy calculation module (103) combines the preset weight vector of each of the key quality attributes to perform a weighted summation of the normalized deviation to obtain the comprehensive deviation index.

5. The dynamic data evaluation system for agricultural product brand value according to claim 4, characterized in that, The first computing channel is configured as follows: Within a set time window, obtain the sequence set of the comprehensive deviation index; The probability density function of the comprehensive deviation index is fitted using the kernel density estimation method; The brand consistency entropy is calculated by performing an integral operation on the probability density function. The brand consistency entropy is used to quantify the uncertainty of a brand's physical quality relative to its promises during the delivery process.

6. The dynamic data evaluation system for agricultural product brand value according to claim 1, characterized in that, The second computing channel is configured as follows: Statistical features of the physical state data of the non-branded agricultural products are obtained, and a background state vector is constructed, which includes mean and dispersion information. Based on the background state vector, a Monte Carlo simulation is performed to generate a set of virtual background physical state vectors; Calculate a background deviation sample set relative to the committed feature vector for the set of virtual background physical state vectors; An environmental background probability distribution is constructed based on the background deviation sample set, and the environmental background entropy is calculated by integrating the environmental background probability distribution.

7. The dynamic data evaluation system for agricultural product brand value according to claim 1, characterized in that, The entropy difference value assessment module (104) is configured as follows when generating the dynamic brand value index: The difference between the environmental background entropy and the brand consistency entropy is calculated as the original entropy difference component, and an environmental pressure compensation factor is introduced to calculate the entropy difference gain. The environmental pressure compensation factor is configured as follows: When the environmental background entropy value is large, the environmental pressure compensation factor approaches one, thereby preserving the original entropy difference component; When the environmental background entropy value is small, the environmental pressure compensation factor approaches zero, thereby suppressing the original entropy difference component.

8. The dynamic data evaluation system for agricultural product brand value according to claim 7, characterized in that, The entropy difference value assessment module (104) is also configured as follows: The historical cumulative credit score is read and time decay is processed using an exponential decay model. The entropy difference gain obtained in this calculation is superimposed on the decayed historical score to obtain the current cumulative credit score. The dynamic brand value index is generated by mapping the cumulative reputation score at the current moment to a closed interval using a logistic function.

9. The dynamic data evaluation system for agricultural product brand value according to claim 1, characterized in that, The dynamic update and storage terminal (105) is configured to add the received brand value index to the historical evaluation data to update the brand value index sequence and construct the time series value ledger that cannot be tampered with. The dynamic update and storage terminal (105) adopts an event-driven mechanism. It triggers persistent storage operation only when the difference between the brand value index sequence and the previous stored value exceeds a preset threshold, or when the time since the last update exceeds the maximum silent period, and outputs the current brand value assessment result to the outside world through the standard data interface.

10. A dynamic data evaluation method for the brand value of agricultural products, characterized in that, The application of the dynamic data evaluation system for agricultural product brand value according to any one of claims 1-9 includes the following steps: Step S100: Establish a dual data monitoring domain for the target and background, and perform spatiotemporal data alignment to generate structured target product physical state data and background physical state data. Step S200: Parse the publicly available marketing text data, construct a marketing commitment feature vector containing physical parameter constraints, and perform semantic anchoring; Step S300: Generate a physical delivery state vector and a background state vector, which respectively represent the real-time physical value of the target product and the industry average physical state statistical characteristics under the environmental background. Step S400: Calculate the dynamic deviation index. By comparing the physical delivery status vector with the marketing commitment feature vector, quantify the performance deviation of the target brand in each attribute. Step S500: Perform dual-track entropy calculation, construct the probability density function of the deviation index to calculate the brand consistency entropy, and construct the background distribution model to calculate the environmental background entropy. Step S600: Generate entropy difference gain and brand value index based on entropy difference model. By calculating the difference between environmental background entropy and brand consistency entropy, and combining nonlinear compensation model and time series accumulation model, generate dynamic brand value index at the current moment. Step S700: Perform a time-series update of the brand value, using a smoothing algorithm to update the brand value sequence and store it in the system database.

Images