Food production efficiency prediction method based on big data processing
By constructing a hygiene risk map and graph neural network to optimize food production efficiency prediction, the problem that existing methods do not consider hygiene standards is solved, a balance between hygiene compliance and efficiency in the food production process is achieved, and the accuracy and compliance of the prediction results are ensured.
Patent Information
- Application Number
- CN202510849894.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-24
- Publication Date
- 2025-09-23
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing production efficiency prediction methods fail to effectively consider the hygiene standards of the food industry, resulting in significant deviations between the prediction results and the actual situation.
By constructing a hygiene risk map including equipment nodes, product nodes and cleaning method nodes, using a preset graph neural network, training the graph neural network based on production big data, calculating the equipment hygiene characteristics and correcting the initial production efficiency data, and combining the cross-contamination risk index to optimize the production efficiency data.
It achieves an intelligent balance between production capacity forecasting and hygiene compliance, avoids efficiency loss caused by excessive cleaning, prevents production hazards caused by hygiene risks, and ensures the accuracy and compliance of forecast results.
Smart Images
Figure CN120688692A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of production management, and in particular to a food production efficiency prediction method based on big data processing. Background Art
[0002] Production efficiency forecasting is crucial in modern industry, helping companies optimize resource allocation, reduce operating costs, and enhance market competitiveness. By accurately forecasting production performance, companies can rationalize production plans, reduce equipment idle time, and dynamically adjust supply chain management, ultimately achieving efficient and sustainable production models.
[0003] However, unlike other sectors, the food industry requires not only efficiency but also compliance with stringent hygiene standards, which often take precedence over factors like efficiency. Existing production efficiency prediction technologies are often based on general industrial data modeling, often overlooking the food industry's mandatory requirements for hygiene compliance. For example, existing methods may only predict production based on parameters such as historical production capacity and equipment status, without incorporating hygiene-related factors such as cleaning cycles and disinfection frequency. This can lead to significant deviations between predicted results and actual production conditions.
[0004] Therefore, people need a food production efficiency prediction method based on big data processing that can take into account hygiene standards. Summary of the Invention
[0005] The purpose of this invention is to provide a food production efficiency prediction method based on big data processing to solve the following technical problems: Existing production efficiency prediction methods do not consider the impact of hygiene standards.
[0006] The purpose of the present invention can be achieved through the following technical solutions: A food production efficiency prediction method based on big data processing includes the following steps: Obtain production environment data, production plan data, and cleaning data; Based on the production environment data, the initial production efficiency data is obtained; Based on production plan data and cleaning data, a hygiene risk graph is established. The nodes in the hygiene risk graph include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attributes corresponding to the node. The edges in the hygiene risk graph represent interactive relationships. Based on the hygiene risk map, the hygiene characteristics of each equipment node are obtained based on a preset graph neural network, where the preset graph neural network is built based on production big data; The initial production efficiency data is optimized according to the hygiene characteristics to obtain the target production efficiency data.
[0007] As a further solution of the present invention: obtaining initial production efficiency data based on production environment data, including: Establish an environment description vector based on production environment data; Input the environmental description vector into a preset neural network model to obtain the production efficiency prediction data for a period of time in the future output by the preset neural network model; According to the production environment data and based on the theoretical mathematical model, the theoretical production efficiency data is obtained; According to the theoretical production efficiency data, the production efficiency prediction data is corrected to obtain the initial production efficiency data.
[0008] As a further solution of the present invention: the theoretical production efficiency data includes the theoretical maximum value of the production efficiency data; based on the theoretical production efficiency data, the production efficiency prediction data is corrected to obtain the initial production efficiency data, including: Compare the production efficiency forecast data with the theoretical maximum production efficiency data; The data exceeding the theoretical maximum value in the production efficiency forecast data are replaced with the theoretical maximum value.
[0009] As a further solution of the present invention: the edges in the hygiene risk graph include edges connecting equipment nodes and product nodes, edges connecting equipment nodes and cleaning method nodes, and edges connecting product nodes and cleaning method nodes; the lengths of the feature vectors corresponding to the equipment nodes, product nodes, and cleaning method nodes are the same, each feature vector includes a type element representing the node type, and the position of the type element in different feature vectors is the same.
[0010] As a further embodiment of the present invention, the hygiene characteristics include a cross-contamination risk index; and the food production efficiency prediction method based on big data processing further includes: Obtain production big data; Establish input health risk maps based on production big data; Based on the production big data, the equipment that has experienced cross contamination in the input health risk map is obtained, and the output data of the node corresponding to the equipment in the input health risk map is recorded as the maximum value of the cross contamination risk index. The output data of the node corresponding to the equipment directly connected to the equipment in the input health risk map is recorded as the median value of the cross contamination risk index to obtain training data; Train a preset graph neural network based on the training data.
[0011] As a further solution of the present invention, the hygiene characteristics include a cross-contamination risk index; the production efficiency data includes an initial predicted output; the initial production efficiency data is optimized according to the hygiene characteristics to obtain target production efficiency data, including: According to the cross contamination risk index, the amount of cross contamination scrap is obtained; Get the standard cleaning time and get the actual cleaning time based on the cross-contamination risk index; According to the actual cleaning time, the cleaning loss output is obtained; The initial predicted output is corrected according to the cross-contamination scrap volume and the cleaning loss output to obtain the actual predicted output as the target production efficiency data.
[0012] As a further solution of the present invention, the hygienic characteristics include a cross-contamination risk index; the production efficiency data include an initial equipment overall efficiency; the initial production efficiency data is optimized according to the hygienic characteristics to obtain target production efficiency data, including: Get the standard cleaning time and get the actual cleaning time based on the cross-contamination risk index; The proportion of cleaning loss time based on actual cleaning time; According to the proportion of cleaning loss time, the initial equipment overall efficiency is corrected to obtain the actual equipment overall efficiency as the target production efficiency data.
[0013] A food production efficiency prediction system based on big data processing, comprising: Data collection module, used to obtain production environment data, production plan data and cleaning data; The primary prediction module is used to obtain initial production efficiency data based on production environment data; The environmental analysis module is used to create a hygiene risk map based on production plan data and cleaning data. The nodes in the hygiene risk map include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attributes corresponding to the node. The edges in the hygiene risk map represent interactive relationships. A risk analysis module is used to obtain the health characteristics of each equipment node based on the health risk map and a preset graph neural network, wherein the preset graph neural network is built based on production big data; The efficiency correction module is used to optimize the initial production efficiency data according to the hygiene characteristics to obtain the target production efficiency data.
[0014] An electronic device, comprising: memory and processor; The memory is used to store the program, and the processor is used to perform the steps of any of the above-mentioned food production efficiency prediction methods based on big data processing when executing the program.
[0015] A computer-readable storage medium for storing computer-readable programs or instructions, which, when executed by a processor, can implement the steps of any of the above-mentioned food production efficiency prediction methods based on big data processing.
[0016] Beneficial effects of the present invention: The present invention provides a food production efficiency prediction method based on big data processing. The method first obtains production environment data, production plan data, and cleaning data, obtains initial production efficiency data based on the production environment data, and then establishes a hygiene risk graph based on the production plan data and the cleaning data. The nodes in the hygiene risk graph include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attribute corresponding to the node. The edges in the hygiene risk graph represent the interaction relationship. Then, based on the hygiene risk graph, the hygiene characteristics of each equipment node are obtained based on a preset graph neural network, wherein the preset graph neural network is established based on production big data. Finally, the initial production efficiency data is optimized based on the hygiene characteristics to obtain target production efficiency data. By constructing a hygiene risk graph including equipment nodes, product nodes, and cleaning method nodes, and using feature vectors to represent the attributes of each node, the present invention can accurately depict the complex hygiene interaction relationships in the food production process, ensuring the model's ability to model hygiene risk factors. In addition, the present invention uses a graph neural network trained based on production big data to calculate the equipment hygiene characteristics and correct the initial production efficiency data, making full use of data-driven learning capabilities to make the extraction of hygiene characteristics more objective and accurate, and achieve an intelligent balance between production capacity prediction and hygiene compliance. It not only avoids efficiency losses caused by excessive cleaning, but also prevents production hazards caused by hygiene risks, and solves the problem that existing production efficiency prediction methods do not consider the impact of hygiene standards. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The present invention will be further described below with reference to the accompanying drawings.
[0018] Figure 1 is a method flow chart of the food production efficiency prediction method based on big data processing of the present invention; Figure 2 for Figure 1 Specific step diagram of step S102; Figure 3 This is a system architecture diagram of the food production efficiency prediction system based on big data processing of the present invention. DETAILED DESCRIPTION
[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.
[0020] See also Figure 1 As shown, the present invention is a food production efficiency prediction method based on big data processing, comprising the following steps: S101. Acquire production environment data, production plan data, and cleaning data; S102. Obtaining initial production efficiency data based on production environment data; S103. Create a hygiene risk graph based on the production plan data and the cleaning data. The nodes in the hygiene risk graph include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attributes corresponding to the node. The edges in the hygiene risk graph represent interactive relationships. S104. Obtaining the hygiene characteristics of each equipment node based on the hygiene risk map and a preset graph neural network, wherein the preset graph neural network is established based on production big data; S105. Optimize the initial production efficiency data according to the hygiene characteristics to obtain target production efficiency data.
[0021] Production environment data refers to any data that can be used to predict production efficiency using existing technologies, including equipment data (such as mixers, sterilizers, equipment materials, manufacturers, and installation dates), product data (such as product categories, cost and selling prices), resource data (such as raw material types and supply cycles), environmental data (such as temperature, humidity, and airborne bacterial counts in the workshop), and process data (such as mixing parameters, fermentation time, and pH). Production planning data refers to more specific data required for production processes, such as order details, production schedules, and process routes. Cleaning data refers to data related to the cleaning methods currently available in the production process, including cleaning methods (CIP automatic cleaning, manual wiping), cleaning time (minutes), disinfectant concentration (%), and post-cleaning ATP value (microbial residue). Production efficiency data refers to data required to represent production efficiency based on actual conditions, such as output, overall equipment effectiveness (OEE), and capacity utilization, and can be flexibly set based on actual conditions. Hygiene characteristics refer to data related to hygiene standards that can be used to correct production efficiency data, such as cleaning frequency, cleaning time, contamination risk level, cleaning cost, etc. Similarly, the specific types of hygiene characteristics can also be set according to specific needs.
[0022] By constructing a hygiene risk graph containing equipment nodes, product nodes, and cleaning method nodes, and using feature vectors to represent the attributes of each node, the present invention can accurately depict the complex hygiene interactions in the food production process, ensuring the model's ability to model hygiene risk factors. Furthermore, the present invention uses a graph neural network trained based on production big data to calculate equipment hygiene characteristics and correct initial production efficiency data, fully leveraging data-driven learning capabilities to make the extraction of hygiene characteristics more objective and accurate, achieving an intelligent balance between production capacity prediction and hygiene compliance, avoiding both efficiency losses caused by excessive cleaning and production risks caused by hygiene risks, and resolving the problem that existing production efficiency prediction methods do not consider the impact of hygiene standards.
[0023] Initial production efficiency data refers to production efficiency data obtained by any existing means (data obtained without considering hygiene standards). Similarly, step S102 of the present invention, obtaining initial production efficiency data based on production environment data, can also be achieved by any existing method. However, the present invention also provides a more preferred method for obtaining initial production efficiency data: Combine Figure 2 As shown, step S102 in this embodiment, obtaining initial production efficiency data based on production environment data, specifically includes: S201. Establish an environment description vector based on production environment data; S202: Input the environment description vector into a preset neural network model to obtain production efficiency prediction data for a future period output by the preset neural network model; S203. Obtain theoretical production efficiency data based on the production environment data and a theoretical mathematical model; S204: According to the theoretical production efficiency data, the production efficiency prediction data is corrected to obtain initial production efficiency data.
[0024] In the above process, the environment description vector encodes the current production environment and can include numerical representations of the following data: production planning data (order demand, schedule), theoretical equipment capacity (such as the maximum hourly output specified in the equipment manual), raw material availability (available inventory after deducting safety stock), staffing (number of shift workers, operating efficiency), etc. The pre-set neural network can be any existing network model capable of predicting production efficiency data, such as an LSTM model. This model can capture seasonal fluctuations (such as holiday demand surges) and trend changes (such as capacity ramp-up) based on historical production data for the same period (e.g., the same month or week), and output a baseline production forecast (tons / hour) for the next seven days. The theoretical mathematical model, ideally, uses physical modeling and mathematical derivation to derive theoretical production efficiency data from production environment data.
[0025] This embodiment uses multimodal data fusion and model collaborative optimization to first convert production environment data (such as temperature, humidity, equipment status, etc.) into a high-dimensional environment description vector, and input it into a preset neural network model (such as LSTM or Transformer) for time series prediction to obtain a data-driven future production efficiency prediction value. Combined with a theoretical mathematical model (such as a production capacity calculation model based on physical laws or empirical formulas), the theoretical production efficiency benchmark value is calculated. By dynamically calibrating the neural network prediction results with the theoretical values, it effectively compensates for the possible overfitting or insufficient generalization problems of pure data-driven models, as well as the defect of traditional theoretical models in responding slowly to real-time environmental changes.
[0026] Specifically, in a preferred embodiment, the theoretical production efficiency data includes the theoretical maximum value of the production efficiency data. On this basis, the above step S204, based on the theoretical production efficiency data, corrects the production efficiency prediction data to obtain the initial production efficiency data, specifically including: Compare the production efficiency forecast data with the theoretical maximum production efficiency data; The data exceeding the theoretical maximum value in the production efficiency forecast data are replaced with the theoretical maximum value.
[0027] This embodiment uses the theoretical maximum value of production efficiency as a baseline constraint. By comparing and verifying the production efficiency data predicted by the neural network against the theoretical maximum value, it can effectively identify and correct possible prediction outliers (such as predictions that exceed physical limits due to data noise or model overfitting). This efficient correction is achieved through a simple threshold truncation operation, with low computational cost and easy engineering implementation. In particular, in industries sensitive to production capacity fluctuations, such as food production, this method can prevent over-scheduling of resources due to inflated predictions while ensuring maximum production efficiency within a reasonable range, achieving a perfect balance between prediction accuracy and production feasibility.
[0028] Furthermore, in a preferred embodiment, the edges in the hygiene risk graph include edges connecting device nodes and product nodes, edges connecting device nodes and cleaning method nodes, and edges connecting product nodes and cleaning method nodes; the lengths of the feature vectors corresponding to the device nodes, product nodes, and cleaning method nodes are the same, each feature vector includes a type element representing the node type, and the position of the type element in different feature vectors is the same.
[0029] Specifically, in this embodiment, edges are only set between different types of nodes. The edge between the device node and the product node indicates that the product requires the device for production and processing. The edge between the product node and the cleaning method node indicates the cleaning method to be adopted after the product is produced. The edge between the device node and the cleaning method node indicates that the device has used the cleaning method.
[0030] On the other hand, graph neural networks require that the feature vectors of all nodes in the graph have the same length in order to perform matrix operations. Therefore, in this embodiment, a type element is added to the feature vector to reflect the differences between different node types. For example, the feature vector of a device node can include data such as the device ID, material, manufacturer, most recently produced product category, and ATP value after the last cleaning. The feature vector of a product node can include product ID, category name, microbial sensitivity level (e.g., dairy products = level 3, baked goods = level 1), etc. The feature vector of a cleaning method can include cleaning method ID, cleaning time, disinfectant concentration, microbial kill efficiency, etc. Among them, the device ID, product ID, and cleaning method ID are the above-mentioned type elements.
[0031] The graph structure design in this embodiment fully captures the critical hygiene path in the food production process through a linkage mechanism between three different types of nodes. By forcing a uniform feature vector length and innovatively introducing type elements (such as device ID, product ID, and cleaning method ID) as fixed position identifiers, this approach not only meets the matrix computation requirements of graph neural networks (GNNs) but also enables explicit differentiation of node types. This allows the characteristics of different node types (such as the ATP value of a device, the microbial sensitivity level of a product, and the disinfectant concentration of a cleaning method) to be integrated and calculated within the same framework, ensuring the standardization of model inputs while preserving the specialized attributes of each node.
[0032] Furthermore, in a preferred embodiment, the hygiene characteristics include a cross-contamination risk index, which is used to indicate the probability of microbial contamination caused by incomplete cleaning or historical contamination when the equipment produces the next batch of products, so that the cleaning time can be inferred based on this, and the production efficiency can be corrected. However, it is understandable that in practice, the cross-contamination risk index is not data that can be directly collected, so when training the graph neural network, it is necessary to quantify the relevant data to represent the cross-contamination risk index. Therefore, in this embodiment, the food production efficiency prediction method based on big data processing also includes the following steps: Obtain production big data; Establish input health risk maps based on production big data; Based on the production big data, the equipment that has experienced cross contamination in the input health risk map is obtained, and the output data of the node corresponding to the equipment in the input health risk map is recorded as the maximum value of the cross contamination risk index. The output data of the node corresponding to the equipment directly connected to the equipment in the input health risk map is recorded as the median value of the cross contamination risk index to obtain training data; Train a preset graph neural network based on the training data.
[0033] For example, a cross-contamination risk index output by a pre-set graph neural network is a value between 0 and 1, where 0 indicates no cross-contamination risk at all and 1 indicates a definite cross-contamination risk. Therefore, in the training data, the node corresponding to a device experiencing cross-contamination in the input hygiene risk graph will have an output value of 1, while the node corresponding to a device directly connected to that device will have an output value of 0.5.
[0034] This embodiment is based on the reverse reasoning mechanism of real cross-contamination events. The device node where the actual contamination occurs is marked as the maximum risk value 1, and the device node directly associated with it is marked as the median value 0.5. This propagation-based labeling method based on the graph structure not only conforms to the actual propagation law of microbial contamination in food production, but also realizes the quantitative prediction of risk value through the graph neural network. The cross-contamination risk that cannot be directly observed is converted into a computable graph neural network training target, so that the model can learn the topological characteristics of contamination propagation and the implicit correlation law between devices. Compared with the traditional method that relies on manual experience for labeling, the prediction results are more objective and accurate. More importantly, this hierarchical labeling strategy not only retains the severity information of the contamination incident, but also retains the characteristics of the contamination propagation path through the graph structure, so that the model can capture local contamination incidents and systemic health risks at the same time.
[0035] Furthermore, in a preferred embodiment, the hygiene characteristics include a cross-contamination risk index, and the production efficiency data includes an initial predicted output. On this basis, the above step S105, optimizing the initial production efficiency data according to the hygiene characteristics to obtain the target production efficiency data, specifically includes: According to the cross contamination risk index, the amount of cross contamination scrap is obtained; Get the standard cleaning time and get the actual cleaning time based on the cross-contamination risk index; According to the actual cleaning time, the cleaning loss output is obtained; The initial predicted output is corrected according to the cross-contamination scrap volume and the cleaning loss output to obtain the actual predicted output as the target production efficiency data.
[0036] In the above process, the amount of cross-contamination scrap can be obtained by multiplying the scrap rate corresponding to the cross-contamination risk index (a preset value obtained based on experience or experiments, such as cross-contamination risk index ≥ 0.7, scrap rate 5%) by the planned output for the day.
[0037] The actual cleaning time can be obtained by the following formula: ; Where T is the actual cleaning time, t is the standard cleaning time, R is the cross-contamination risk index, and δ is the preset risk adjustment coefficient.
[0038] The cleaning loss output can be obtained by multiplying the sum of the actual cleaning time of all equipment per day by the theoretical production capacity during that period.
[0039] This embodiment introduces the cross-contamination risk index as a core hygiene feature. It not only calculates the amount of scrap loss directly caused by it, but also innovatively combines the standard cleaning time and the risk index to deduce the actual cleaning time required, and then quantifies the production capacity loss caused by the cleaning process. Ultimately, a more realistic and reliable target output forecast is obtained through the dual correction mechanism of "scrap amount + cleaning loss amount", which converts the hygiene risk into a calculable production capacity loss indicator, and realizes the organic unity of hygiene management and production efficiency. It avoids both the waste of production capacity caused by excessive cleaning and the quality accidents caused by substandard hygiene.
[0040] Furthermore, in a preferred embodiment, the hygiene characteristics include a cross-contamination risk index, and the production efficiency data include an initial overall equipment efficiency (OEE). On this basis, the above step S105, optimizing the initial production efficiency data according to the hygiene characteristics to obtain target production efficiency data, specifically includes: Get the standard cleaning time and get the actual cleaning time based on the cross-contamination risk index; The proportion of cleaning loss time based on actual cleaning time; According to the proportion of cleaning loss time, the initial equipment overall efficiency is corrected to obtain the actual equipment overall efficiency as the target production efficiency data.
[0041] The above process can be expressed by the following formula: ; Among them, O is the actual equipment comprehensive efficiency, oee is the initial equipment comprehensive efficiency, and r is the proportion of cleaning loss time.
[0042] This example dynamically infers actual cleaning time through the cross-contamination risk index, accurately calculating the proportion of cleaning loss time, and ultimately achieving a scientific correction to the initial OEE value. A dynamic correlation model between hygiene risk and equipment efficiency has been established, making OEE assessments more consistent with actual food production conditions. This not only reflects the additional cleaning time required after producing high-risk products, but also reflects the differences in hygiene management between different equipment. This enhances the applicability of OEE data in the food industry and makes production efficiency assessments more comprehensive and accurate.
[0043] Combine Figure 3 As shown, the present invention also provides a food production efficiency prediction system based on big data processing, comprising: Data collection module 310, for acquiring production environment data, production plan data, and cleaning data; A primary prediction module 320 is used to obtain initial production efficiency data based on production environment data; Environmental analysis module 330, for creating a hygiene risk map based on production plan data and cleaning data. Nodes in the hygiene risk map include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attributes corresponding to the node. Edges in the hygiene risk map represent interactions. A risk analysis module 340 is configured to obtain the health characteristics of each equipment node based on the health risk map and a preset graph neural network, wherein the preset graph neural network is established based on production big data; The efficiency correction module 350 is used to optimize the initial production efficiency data according to the hygiene characteristics to obtain the target production efficiency data.
[0044] The present invention further provides an electronic device, comprising: memory and processor; The memory is used to store the program, and the processor is used to perform the steps of any of the above-mentioned food production efficiency prediction methods based on big data processing when executing the program.
[0045] The present invention also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the program or instructions are executed by a processor, the program or instructions can implement any step in the above-mentioned food production efficiency prediction method based on big data processing.
[0046] The present invention provides a food production efficiency prediction method based on big data processing. The method first obtains production environment data, production plan data, and cleaning data, obtains initial production efficiency data based on the production environment data, and then establishes a hygiene risk graph based on the production plan data and the cleaning data. The nodes in the hygiene risk graph include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attribute corresponding to the node. The edges in the hygiene risk graph represent the interaction relationship. Then, based on the hygiene risk graph, the hygiene characteristics of each equipment node are obtained based on a preset graph neural network, wherein the preset graph neural network is established based on production big data. Finally, the initial production efficiency data is optimized based on the hygiene characteristics to obtain target production efficiency data. By constructing a hygiene risk graph including equipment nodes, product nodes, and cleaning method nodes, and using feature vectors to represent the attributes of each node, the present invention can accurately depict the complex hygiene interaction relationships in the food production process, ensuring the model's ability to model hygiene risk factors. In addition, the present invention uses a graph neural network trained based on production big data to calculate the equipment hygiene characteristics and correct the initial production efficiency data, making full use of data-driven learning capabilities to make the extraction of hygiene characteristics more objective and accurate, and achieve an intelligent balance between production capacity prediction and hygiene compliance. It not only avoids efficiency losses caused by excessive cleaning, but also prevents production hazards caused by hygiene risks, and solves the problem that existing production efficiency prediction methods do not consider the impact of hygiene standards.
[0047] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.
Claims
1. A food production efficiency prediction method based on big data processing, characterized in that: The following steps are involved: Obtain production environment data, production plan data, and cleaning data; Based on the production environment data, the initial production efficiency data is obtained; Based on production plan data and cleaning data, a hygiene risk graph is established. The nodes in the hygiene risk graph include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attributes corresponding to the node. The edges in the hygiene risk graph represent interactive relationships. Based on the hygiene risk map, the hygiene characteristics of each equipment node are obtained based on a preset graph neural network, where the preset graph neural network is built based on production big data; The initial production efficiency data is optimized according to the hygiene characteristics to obtain the target production efficiency data.
2. The food production efficiency prediction method based on big data processing according to claim 1 is characterized in that: Based on the production environment data, the initial production efficiency data is obtained, including: Establish an environment description vector based on production environment data; Input the environmental description vector into a preset neural network model to obtain the production efficiency prediction data for a period of time in the future output by the preset neural network model; According to the production environment data and based on the theoretical mathematical model, the theoretical production efficiency data is obtained; According to the theoretical production efficiency data, the production efficiency prediction data is corrected to obtain the initial production efficiency data.
3. The food production efficiency prediction method based on big data processing according to claim 2 is characterized in that: Theoretical production efficiency data includes the theoretical maximum value of production efficiency data; According to the theoretical production efficiency data, the production efficiency prediction data is corrected to obtain the initial production efficiency data, including: Compare the production efficiency forecast data with the theoretical maximum production efficiency data; The data exceeding the theoretical maximum value in the production efficiency forecast data are replaced with the theoretical maximum value.
4. The food production efficiency prediction method based on big data processing according to claim 1 is characterized in that: The edges in the hygiene risk graph include edges connecting equipment nodes and product nodes, edges connecting equipment nodes and cleaning method nodes, and edges connecting product nodes and cleaning method nodes; the lengths of the feature vectors corresponding to equipment nodes, product nodes, and cleaning method nodes are the same, each feature vector includes a type element representing the node type, and the position of the type element in different feature vectors is the same.
5. The food production efficiency prediction method based on big data processing according to claim 1 is characterized in that: The hygiene characteristics include a cross-contamination risk index; and the food production efficiency prediction method based on big data processing further includes: Obtain production big data; Establish input health risk maps based on production big data; Based on the production big data, the equipment that has experienced cross contamination in the input health risk map is obtained, and the output data of the node corresponding to the equipment in the input health risk map is recorded as the maximum value of the cross contamination risk index. The output data of the node corresponding to the equipment directly connected to the equipment in the input health risk map is recorded as the median value of the cross contamination risk index to obtain training data; Train a preset graph neural network based on the training data.
6. The food production efficiency prediction method based on big data processing according to claim 1 is characterized in that: Hygiene characteristics include cross-contamination risk index; Production efficiency data include initial forecast output; Optimize the initial production efficiency data according to the hygiene characteristics to obtain the target production efficiency data, including: According to the cross contamination risk index, the amount of cross contamination scrap is obtained; Get the standard cleaning time and get the actual cleaning time based on the cross-contamination risk index; According to the actual cleaning time, the cleaning loss output is obtained; The initial predicted output is corrected according to the cross-contamination scrap volume and the cleaning loss output to obtain the actual predicted output as the target production efficiency data.
7. The food production efficiency prediction method based on big data processing according to claim 1 is characterized in that: Hygiene characteristics include cross-contamination risk index; Production efficiency data include initial overall equipment efficiency; Optimize the initial production efficiency data according to the hygiene characteristics to obtain the target production efficiency data, including: Get the standard cleaning time and get the actual cleaning time based on the cross-contamination risk index; The proportion of cleaning loss time based on actual cleaning time; According to the proportion of cleaning loss time, the initial equipment overall efficiency is corrected to obtain the actual equipment overall efficiency as the target production efficiency data.
8. A food production efficiency prediction system based on big data processing, characterized in that: include: Data collection module, used to obtain production environment data, production plan data and cleaning data; The primary prediction module is used to obtain initial production efficiency data based on production environment data; The environmental analysis module is used to create a hygiene risk map based on production plan data and cleaning data. The nodes in the hygiene risk map include equipment nodes, product nodes, and cleaning method nodes. Each node corresponds to a feature vector representing the object attributes corresponding to the node. The edges in the hygiene risk map represent interactive relationships. A risk analysis module is used to obtain the health characteristics of each equipment node based on the health risk map and a preset graph neural network, wherein the preset graph neural network is built based on production big data; The efficiency correction module is used to optimize the initial production efficiency data according to the hygiene characteristics to obtain the target production efficiency data.
9. An electronic device, characterized in that: include: memory and processor; The memory is used to store the program, and the processor is used to perform the steps of any one of the food production efficiency prediction methods based on big data processing in claims 1-7 when executing the program.
10. A computer-readable storage medium, characterized in that Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps of any one of claims 1-7 in the method for predicting food production efficiency based on big data processing.