A method and system for detecting and analyzing the production quality of goat milk powder

Abstract

The present invention relates to the technical field of quality detection, and in particular to a method and system for detecting and analyzing the production quality of goat milk powder, the method comprising: constructing a parameter sequence according to the actual parameter value of the current process within a preset time period; using the parameter sequence to train a preset network prediction model to obtain a trained network prediction model; inputting the parameter sequence obtained in real time into the trained network prediction model to obtain the first quality prediction value of the next moment in the production process of the process; in response to the first quality prediction value of the next moment being greater than a preset standard threshold, adjusting the actual parameter value of the next moment in the production process of the current process. That is, the scheme of the present invention can calculate the comprehensive quality value of the products of each process, can perform real-time detection and analysis on the production quality in each process, and is convenient for monitoring the quality of products in different processes.

Description

Technical Field

[0001] The present invention relates to the technical field of quality detection, and more specifically, to a method and system for detecting and analyzing the production quality of goat milk powder. Background Art

[0002] Goat milk powder is usually used to make infant formula, adult nutritional products, dairy products, baked goods, etc. It is rich in protein, fat, vitamins and minerals and can be used as a nutritional supplement. In the production process, goat milk powder needs to go through multiple processing steps. The most common processing steps are generally divided into pre-treatment, concentration and deodorization, secondary drying and internal and external packaging. In order to ensure that the product meets safety standards and ensure product quality stability, goat milk powder needs to be tested and analyzed in each production process to ensure the production quality of goat milk powder.

[0003] The Chinese patent application document with publication number CN110796389A discloses a milk powder production quality control analysis system and method, which obtains the formula of the milk powder to be produced, calls the content of the nutrient components of the available batches of materials corresponding to the formula, and calculates the deduction result of the nutrient components of the milk powder to be produced according to the content of the nutrient components. The nutrient component content of the milk powder is deduced, thereby achieving control and analysis of the production quality of the milk powder.

[0004] The above method calculates and deduce the content of nutrient components of available batches of materials corresponding to the formula of the milk powder to be produced, so as to realize the control and analysis of production quality. However, the above related scheme does not monitor and conduct quality inspection on each production process during the production process, and it is impossible to adjust the production parameters in real time to improve the production quality of goat milk powder. Summary of the invention

[0005] The invention proposes a goat milk powder production quality detection and analysis method and system, which are used to solve the problem in the prior art that quality detection cannot be performed on each production process to improve the production quality of milk powder.

[0006] In a first aspect, the present invention provides a method for detecting and analyzing the production quality of goat milk powder, comprising the following steps:

[0007] Construct a parameter sequence based on multiple actual parameter values ​​of the current process within a preset time period;

[0008] Using the parameter sequence to train the preset network prediction model to obtain a trained network prediction model;

[0009] Inputting the parameter sequence obtained in real time into the trained network prediction model to obtain the first quality prediction value at the next moment in the production process of the process;

[0010] In response to the first quality prediction value at the next moment being greater than a preset standard threshold, adjusting the actual parameter value at the next moment in the production process of the current process;

[0011] Among them, the adjustment method of the actual parameter value at the next moment is:

[0012] Select a value from the range of theoretical parameter values ​​as the adjustment parameter;

[0013] Add the adjustment parameters to the parameter sequence to obtain an updated parameter sequence;

[0014] Inputting the updated parameter sequence into a preset time series network model to obtain a second quality prediction value at the next moment;

[0015] Iteratively optimize the adjustment parameters until the difference between the second quality prediction value and the target value is less than a preset threshold, and use the iteratively optimized adjustment parameters as the actual parameter values ​​at the next moment, wherein the target value is negatively correlated with the error value of the first quality prediction value.

[0016] By predicting the first quality prediction value at the next moment, the actual parameter value at the next moment is adjusted in real time according to the first quality prediction value, thereby improving the production quality of the product. In the process of adjusting the actual parameter value, the accuracy of the actual parameter value at the next moment is increased by considering the prediction error of the network prediction model, thereby obtaining a better actual parameter value.

[0017] Preferably, the parameter sequence also includes the comprehensive quality value of the product after the previous process is completed, and the calculation method of the comprehensive quality value of the product is:

[0018]

[0019] in: Indicates The comprehensive quality value of the product when each production process is completed; Indicates The actual parameter value collected at each moment; Indicates Theoretical parameter values ​​set at each moment; Indicated in The number of actual parameter values ​​in a production process, Indicates The processing time of each production process.

[0020] By calculating the comprehensive quality value of the product, the product quality after the process is completed can be intuitively detected and analyzed based on the comprehensive quality value of the product. In the process of the network prediction model predicting the quality value of the next moment of the process, the comprehensive quality value of the product of the previous process is introduced, which improves the accuracy of the prediction results of the time series network model.

[0021] Preferably, the target value is calculated as follows:

[0022]

[0023] in, is the target value, is a pre-set standard threshold, It is the error value of the first quality prediction value predicted by the network prediction model at the current moment.

[0024] In the process of adjusting the actual parameters, the target value is adjusted in real time according to the error of the current timing network model, thereby improving the accuracy of the actual parameter value at the next moment.

[0025] Preferably, the calculation method of the error value of the first quality prediction value predicted at the current moment is:

[0026] Input the actual parameter values ​​at the historical moment into the time series network model to obtain the first quality prediction value at the current moment;

[0027] Calculate the actual quality value at the current moment, and use the difference between the first quality prediction value and the actual quality value at the current moment as the error value of the first quality prediction value;

[0028] Among them: The calculation formula of the actual mass value is:

[0029]

[0030] in: Indicated in In the production process The actual quality value at a moment; Indicates The actual parameter value of the setting collected at each moment; Indicates Theoretical parameter values ​​set at each moment; Indicates the quantity of actual quality value in the current production process. Indicated in From the first production process to the The duration of continuous processing at a moment.

[0031] In the production process, by calculating the error value of the network prediction model at the current moment, it is convenient to optimize the adjustment parameters in real time so that the actual parameters are always in the optimal state, so that the quality value at each moment can reach the optimal value, thereby achieving the purpose of improving production results.

[0032] Preferably, the network prediction model is a long short-term network memory model.

[0033] Preferably, the training method of the network prediction model is:

[0034] Obtain a training set, wherein the training set includes multiple groups of parameter sequences, each group of parameter sequences is provided with a label, each group of parameter sequences includes the actual parameter values ​​of the current process at the previous a different moments in the production process, and the comprehensive quality value of the product after the previous process is completed, wherein the label is the actual quality value at the b-th moment in a preset time period, wherein a is before b;

[0035] The training set is input into the pre-built network model for training. During the training process, the cross entropy loss between the output prediction value and the label is calculated;

[0036] The model parameters are adjusted using the gradient descent method to minimize the error between the predicted value and the actual quality value;

[0037] Iteratively adjust the parameters of the network model until the cross entropy loss is less than a certain value or the set number of training times is reached to obtain the network prediction model.

[0038] Preferably, the method of selecting a value from the value range of the theoretical parameter value as the adjustment parameter is: selecting the maximum value or the minimum value from the value range of the theoretical parameter value as the adjustment parameter.

[0039] By selecting adjustment parameters from a range of values ​​and optimizing the adjustment parameters as actual parameters at the next moment, the processing technology in the production process is improved and the production quality of the product is improved.

[0040] In another aspect, a goat milk powder production quality detection and analysis system comprises:

[0041] processor;

[0042] The memory stores computer instructions for detecting and analyzing the production quality of goat milk powder. When the computer instructions are executed by the processor, the system executes a method for detecting and analyzing the production quality of goat milk powder.

[0043] The beneficial effects of the present invention are:

[0044] The solution of the present invention uses a network prediction model to perform real-time prediction and analysis on the production quality of each process when detecting and analyzing the production quality of goat milk powder, so as to facilitate monitoring of the quality of the product during the production process. At the same time, in the production process of each process, the actual parameters of the next moment are adjusted in real time by referring to the target value, so that the production quality value corresponding to each moment is optimized, so as to achieve the effect of improving the production quality and improving the production quality of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0046] Figure 1 A flowchart of the steps of a method for detecting and analyzing the production quality of goat milk powder in this embodiment is schematically shown;

[0047] Figure 2 The structural block diagram of a goat milk powder production quality detection and analysis system in this embodiment is schematically shown. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0049] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0050] The scenario targeted by the present invention is as follows: fresh goat milk is inspected after entering the factory, and fresh milk that passes the inspection will be transported to the pre-treatment workshop through a pipeline. The milk purifier uses the centrifugal principle to separate impurities in the fresh milk. After cooling, it is sterilized by a pasteurizer, and then enters the batching system for fine batching, and various nutrients are added evenly. After that, the fat globules in the liquid are crushed by a homogenizer, and then transported to the intermediate storage tank for temporary storage. In the concentration workshop, pure steam is directly sprayed into the liquid for sterilization, and then the flash tank is used for deodorization. The liquid concentration is then concentrated to 50% by a three-effect falling film evaporator. In the powder spraying workshop, the concentrated milk preheated by the preheater is transported to the high-pressure spray gun at the top of the tower by a high-pressure pump for uniform atomization, and heat exchange is performed with the filtered and sterilized high-temperature air, so that the atomized concentrated milk is instantly dried into powder, and the milk powder particles fall and separate from the hot air. The spray-dried milk powder is cooled and dried twice in a fluidized bed, and then enters the powder bin, and is packaged into base powder by an automatic packaging machine for use in dry process.

[0051] The present invention is directed to the process of sterilizing fresh goat milk by a pasteurizer and then sterilizing it in a concentration workshop.

[0052] In the first aspect, specifically, Figure 1 As shown, a goat milk powder production quality detection and analysis method in this embodiment comprises the following steps:

[0053] S1. Set the value range of theoretical parameter values ​​for each production process.

[0054] In the production process of goat milk powder, multiple processes are required to produce and process goat milk in sequence. For each production process, the production quality of the product is guaranteed by controlling important production parameters. The theoretical parameter values ​​of each production process are set manually according to the actual production situation.

[0055] For example, during the sterilization process of goat milk powder production, temperature changes directly affect the quality of sterilization. Therefore, in the sterilization production process, temperature is used as an important control parameter.

[0056] During the sterilization process of goat milk powder production, the temperature should not exceed 100 degrees Celsius and should not be lower than 72 degrees Celsius. Therefore, during the sterilization process, the theoretical parameter value for the temperature is within the range of 72-100 degrees Celsius.

[0057] S2. Obtain the actual parameter values ​​in the process, and calculate the comprehensive quality value of the product after the process is completed based on the actual parameter values ​​and theoretical parameter values.

[0058] In this embodiment, the calculation method of the comprehensive quality value of the product after the process is completed is:

[0059]

[0060] in: Indicates The comprehensive quality value of the product when each production process is completed; Indicates The actual parameter value collected at each moment; Indicates Theoretical parameter values ​​set at each moment; Indicated in The number of actual parameter values ​​in a production process, Indicates The actual parameter value is the temperature value.

[0061] For a technological process, the quality of the products of the previous production process will affect the quality of the products of the current production process. After a production process, the comprehensive quality value of the products of the corresponding process is calculated. The comprehensive quality value reflects the quality of the products after the corresponding process is completed. For the comprehensive quality value, the smaller the value, the better the quality of the product.

[0062] Under the condition of stable and normal production, the absolute value of the difference between the theoretical set value and the actual set value of the important control parameters in the production process at the same time is calculated, and the average of multiple absolute values ​​in the production process is calculated as the comprehensive quality value when the current production process is completed.

[0063] Exemplarily, in the sterilization production process, the stable sterilization duration is 10 seconds, and the actual set temperature value is collected at intervals of 1 second. The actual set temperature value is the actual parameter value, and the specific size is set according to the actual situation of the product. The theoretically set temperature value is the theoretical parameter value. For batch-produced products, it is the average value set under theoretical conditions. In the process of sterilizing goat milk powder, different batches of goat milk powder have different quality, and different sterilization times and temperatures are required. For batches with better quality, a lower sterilization temperature and a shorter sterilization time are required. For batches with poorer quality, a higher sterilization temperature and a longer sterilization time are required.

[0064] S3. Construct a parameter sequence according to multiple actual parameter values ​​of the current process within a preset time period; and use the parameter sequence to train a preset network prediction model to obtain a trained network prediction model.

[0065] In this embodiment, the parameter sequence includes the actual parameter values ​​of the current process within a preset time period and the comprehensive quality value of the previous process, wherein the preset time period is all the actual parameter values ​​from the start of the current process to the current moment.

[0066] The method for obtaining the network prediction model is as follows: obtaining a training set, wherein the training set includes multiple groups of data, each group of data includes the actual parameter values ​​of the current process at the previous a different moments in the production process, the comprehensive quality value of the product after the previous process is completed, and a label, wherein the label is the actual quality value at the bth moment in a preset time period, and a is located before b; inputting the training set into a pre-constructed network model for training, and during the training process, calculating the cross entropy loss between the output prediction value and the label; adjusting the model parameters using the gradient descent method to minimize the error between the prediction value and the actual quality value; iteratively adjusting the parameters of the network model until the cross entropy loss is less than a certain value or reaches a set number of training times to obtain the network prediction model, and the network prediction model is a long-short-term network memory model.

[0067] The parameter sequence is input into the preset network prediction model to obtain the quality value of the next moment in the production process of the process.

[0068] For example, the temperature data at the current moment is , the next moment is The temperature data at historical moments are , , the temperature data ( , , ) and the parameter sequence consisting of the comprehensive quality value corresponding to the previous process are input into the network prediction model, and the output is The first quality prediction value at time instant.

[0069] S4. In response to the first quality prediction value at the next moment being greater than the preset standard threshold, the actual parameter value at the next moment in the current production process is adjusted; if the first quality prediction value at the next moment is less than the preset standard threshold, production is performed according to the set theoretical parameter value.

[0070] If the first quality prediction value at the next moment is greater than the preset standard threshold, it means that the production quality corresponding to the next moment is poor, and the actual parameters at the next moment need to be adjusted. If the first quality prediction value at the next moment is less than the preset standard threshold, it means that the production quality corresponding to the next moment is good, and production and processing only need to be carried out according to the set theoretical parameter values.

[0071] Exemplarily, the standard threshold is set based on experience, for example, the standard threshold is 0.36. It can also be determined based on actual conditions. If a higher pass rate is required, the threshold can be set lower.

[0072] The actual parameter value is adjusted as follows:

[0073] S41. Select a value from the range of theoretical parameter values ​​as an adjustment parameter, and add the adjustment parameter to the parameter sequence to obtain an updated parameter sequence.

[0074] For example, the temperature data at the current moment is The temperature data at historical moments are , , adjustment parameters , the comprehensive quality value of the previous process is , then the data sequence is ( , , , , ), adjustment parameters is the maximum or minimum value selected from the temperature range.

[0075] S42: Input the updated parameter sequence into a preset time series network model to obtain a second quality prediction value at the next moment.

[0076] The data series ( , , , , ) is input into the long short-term network memory model to obtain the second quality prediction value at the next moment .

[0077] S43. Obtain target value.

[0078] S431, input the actual parameter value at the historical moment into the time series network model to obtain the first quality prediction value at the current moment, for example, the data sequence ( , , ) is input into the long short-term network memory model to obtain the first quality prediction value at the current moment.

[0079] S432, calculating the actual quality value at the current moment, the calculation formula of the actual quality value is:

[0080]

[0081] in: Indicated in In the production process The actual quality value at a moment; Indicates The actual parameter value of the setting collected at each moment; Indicates Theoretical parameter values ​​set at each moment; Indicates the quantity of actual quality value in the current production process. Indicated in From the first production process to the The duration of continuous processing at a moment.

[0082] For example, the temperature data at the current moment is The temperature data at historical moments are , , the temperature data ( , , ) is input into the calculation formula to obtain the actual mass value at the current moment.

[0083] S433: Taking the difference between the first quality prediction value and the actual quality value at the current moment as the error value of the first quality prediction value.

[0084] By comparing the first quality prediction value and the actual quality value at the current moment, the error value of the first quality prediction value is calculated, and the error value of the first quality is also the prediction error of the time series network model at the current moment.

[0085] S434. Calculate the target value according to the error value of the first quality prediction value.

[0086] In one embodiment, the target value is calculated as follows:

[0087]

[0088] In the formula, is the target value, is the preset standard threshold, is the error value of the first quality prediction value predicted by the network prediction model at the current moment. Among them, the standard threshold Artificially set based on actual experience.

[0089] S435, iteratively optimize the adjustment parameters until the difference between the second quality prediction value and the target value is smaller than a preset threshold, and use the iteratively optimized adjustment parameters as the actual parameter values ​​at the next moment.

[0090] The threshold is set according to actual experience. For example, the threshold is set to 0.02.

[0091] The network prediction model is used to predict and analyze the production quality in each process in real time, which is convenient for monitoring the quality of products in the process of production. At the same time, in the production process of each process, the actual parameters of the next moment are adjusted in real time by referring to the target value, so that the production quality value corresponding to each moment is optimized, thereby improving the production quality and improving the production quality of products.

[0092] Figure 2 The structural block diagram of a goat milk powder production quality detection and analysis system in this embodiment is schematically shown.

[0093] The present invention also provides a goat milk powder production quality detection and analysis system. Figure 2 As shown, the production quality detection and analysis system includes a processor and a memory, and the memory stores computer program instructions. When the computer program instructions are executed by the processor, the above-mentioned goat milk powder production quality detection and analysis method according to the present invention is implemented.

[0094] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0095] The solution of the present invention can detect and analyze the production quality of goat milk powder in real time by calculating the comprehensive quality value of the products in each process when detecting and analyzing the production quality of goat milk powder, which is convenient for monitoring the quality of products in different processes. At the same time, in the production process of each process, the quality value at a future moment is predicted in real time, and the production parameters at the next moment are dynamically adjusted according to the quality value, thereby achieving the effect of improving the production quality and improving the production quality of the product.

[0096] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0097] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for detecting and analyzing the production quality of goat milk powder, characterized in that: The following steps are involved: Construct a parameter sequence based on multiple actual parameter values ​​of the current process within a preset time period and the comprehensive quality value of the product after the previous process; The calculation method of the comprehensive quality value of the product is: ; Indicates The comprehensive quality value of the product when each production process is completed; Indicates The actual parameter value collected at each moment; Indicates Theoretical parameter values ​​set at each moment; Indicated in The number of actual parameter values ​​in a production process, Indicates The processing time of each production process; Using the parameter sequence to train the preset network prediction model to obtain a trained network prediction model; Inputting the parameter sequence obtained in real time into the trained network prediction model to obtain the first quality prediction value at the next moment in the production process of the process; In response to the first quality prediction value at the next moment being greater than a preset standard threshold, adjusting the actual parameter value at the next moment in the production process of the current process; Among them, the adjustment method of the actual parameter value at the next moment is: Select a value from the range of theoretical parameter values ​​as the adjustment parameter; Add the adjustment parameters to the parameter sequence to obtain an updated parameter sequence; Inputting the updated parameter sequence into a preset time series network model to obtain a second quality prediction value at the next moment; Iteratively optimize the adjustment parameters until the difference between the second quality prediction value and the target value is less than a preset threshold, and use the iteratively optimized adjustment parameters as the actual parameter values ​​at the next moment; The calculation method of the error value of the first quality prediction value predicted at the current moment is as follows: the actual parameter value at the historical moment is input into the time series network model to obtain the first quality prediction value at the current moment; the actual quality value at the current moment is calculated, and the difference between the first quality prediction value and the actual quality value at the current moment is used as the error value of the first quality prediction value; the calculation formula of the actual quality value is: ; Indicated in In the production process The actual quality value at a moment; Indicates the quantity of actual quality value in the current production process. Indicated in From the first production process to the The duration of processing at each moment; The target value is calculated as: ; is the target value, is a pre-set standard threshold, It is the error value of the first quality prediction value predicted by the network prediction model at the current moment.

2. A goat milk powder production quality detection and analysis method according to claim 1, characterized in that, The network prediction model is a long short-term network memory model.

3. A goat milk powder production quality detection and analysis method according to claim 1, characterized in that, The training method of the network prediction model is: Obtain a training set, wherein the training set includes multiple groups of parameter sequences, each group of parameter sequences is provided with a label, each group of parameter sequences includes the actual parameter values ​​of the current process at the previous a different moments in the production process, and the comprehensive quality value of the product after the previous process is completed, wherein the label is the actual quality value at the b-th moment in a preset time period, wherein a is before b; The training set is input into the pre-built network model for training. During the training process, the cross entropy loss between the output prediction value and the label is calculated; The model parameters are adjusted using the gradient descent method to minimize the error between the predicted value and the actual quality value; Iteratively adjust the parameters of the network model until the cross entropy loss is less than a certain value or the set number of training times is reached to obtain the network prediction model.

4. A goat milk powder production quality detection and analysis method according to claim 1, characterized in that, The method of selecting a value from the value range of the theoretical parameter value as the adjustment parameter is: selecting the maximum value or the minimum value from the value range of the theoretical parameter value as the adjustment parameter.

5. A goat milk powder production quality detection and analysis system, characterized in that: include: processor; A memory storing computer instructions for goat milk powder production quality detection and analysis. When the computer instructions are executed by the processor, the system executes a goat milk powder production quality detection and analysis method according to any one of claims 1 to 4.

Images