Early warning and control system for milk powder production process

Abstract

The invention belongs to the technical field of production supervision, and discloses an early warning and control system for a milk powder production process, which comprises the steps of acquiring process control points corresponding to each production sub-process in a whole milk powder production process, and performing data acquisition on the corresponding process control points based on a preset data terminal, obtaining corresponding process monitoring data; performing data optimization on the corresponding flow monitoring data, judging whether the corresponding production sub-flow is in a controlled state based on the flow monitoring data after data optimization, and generating corresponding risk early warning information if the corresponding production sub-flow is not in the controlled state; according to the method, risk early warning of different degrees is carried out on the basis of the obtained risk early warning information, corresponding deviation correction instructions are generated, process control is carried out on the corresponding production sub-processes on the basis of the deviation correction instructions, and result feedback is carried out on the corresponding process control results, so that the method is beneficial to optimizing the production processes and improving the resource utilization efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of production supervision, and more specifically, to an early warning and control system for the milk powder production process. Background Art

[0002] The milk powder production process involves multiple links such as raw material procurement, storage, processing, and packaging. Quality risks in any link (such as microbial contamination, veterinary drug residues, equipment failures, etc.) may affect product safety. Traditional milk powder production early warning systems mainly rely on single-parameter threshold monitoring or manual experience judgment, and have the following deficiencies: single-point monitoring is carried out for some key processes, and no full-process risk linkage analysis is formed, making it difficult to discover multi-link superimposed risks (such as the synergistic effect of abnormal raw material storage environment and processing parameter fluctuations); and only simple statistical analysis is carried out on the collected production data, without mining the trend, periodic laws, and noise interference in the data, resulting in insufficient risk prediction accuracy. After risk disposal, there is no effective feedback and optimization mechanism, and it is unable to dynamically adapt to production process changes or new risk types; and the efficiency of manual intervention is low, which is prone to cause production interruptions or quality accidents.

[0003] In view of this, the present invention proposes an early warning and control system for the milk powder production process to solve the above problems. Summary of the Invention

[0004] In order to overcome the above defects of the prior art and to achieve the above object, the present invention provides the following technical solutions: An early warning and control system for the milk powder production process, comprising a process acquisition module, configured to obtain process control points corresponding to each production sub-process within the entire milk powder production process, and based on a pre-set data terminal, perform data acquisition on the corresponding process control points to obtain corresponding process monitoring data; a process analysis module, configured to divide the process monitoring data into a trend sequence, a periodic sequence, and a noise sequence, and based on this, perform data optimization on the corresponding process monitoring data, and based on the process monitoring data after data optimization, determine whether the corresponding production sub-process is in a controlled state. If not, generate corresponding risk warning information; a process warning module, based on the obtained risk warning information, perform risk warnings at different levels and generate corresponding rectification instructions, and based on this, perform process control on the corresponding production sub-process and feed back the corresponding process control results.

[0005] Further, the process of obtaining the process monitoring data includes: Obtain the entire process during the production of the target milk powder and divide it into several production sub-processes; obtain the known reasons for milk powder quality risks, and based on them, conduct a hazard analysis at multiple risk levels for each production sub-process to obtain the process operations that have a direct impact in the corresponding production sub-processes during the milk powder production process; and set them as process control points. Set up data collection terminals, and the data collection terminals collect data from each process control point based on a preset collection cycle to obtain corresponding process monitoring data.

[0006] Furthermore, the process of optimizing the corresponding process monitoring data includes: Obtain the corresponding process monitoring data at each process control point and perform data preprocessing on it. The data preprocessing includes sequence synchronization processing and redundancy processing. Divide the process monitoring data after data preprocessing into a trend sequence, a periodic sequence, and a noise sequence. Obtain the noise mean and noise standard deviation corresponding to the corresponding noise sequence, and based on them, obtain the noise fluctuation coefficient corresponding to each data point in the corresponding process monitoring data; and judge whether the corresponding data point is an abnormal data point based on the noise fluctuation coefficient. If so, perform outlier processing on the corresponding abnormal data point; obtain the corresponding optimized process monitoring data.

[0007] Furthermore, the trend sequence is composed of the corresponding trend estimated values at each data point, which is used to reflect the overall change trend at each data point in the corresponding process monitoring data; the noise sequence is composed of the corresponding noise estimated values at each data point in the corresponding process monitoring data, which is used to reflect the data fluctuation degree at each data point in the corresponding process monitoring data; the periodic sequence is composed of the corresponding periodic estimated values at each data point in the corresponding process monitoring data, which is used to reflect the periodic change law of the corresponding process monitoring data.

[0008] Furthermore, the process of outlier processing includes: Obtain the trend estimated value, noise estimated value, and periodic estimated value corresponding to the data points at adjacent times to the abnormal data point, and based on them, perform data correction on the trend estimated value, noise estimated value, and periodic estimated value at the corresponding abnormal data point. And based on the corrected trend estimated value, noise estimated value, and periodic estimated value, perform data update on the original process monitoring data; obtain the corresponding optimized process monitoring data. The process of data correction includes: Obtain the trend estimation values corresponding to the data points adjacent to the abnormal data points, and combine the linear interpolation algorithm to perform data fitting on the trend estimation values corresponding to the corresponding abnormal data points to obtain the corresponding expected trend estimation values; at the same time, obtain the periodic change rules corresponding to the corresponding periodic sequences, and obtain the corresponding expected periodic estimation values based on the moments at which the corresponding abnormal data points are located in the corresponding periodic change rules; randomly generate an expected noise estimation value according to the mean and standard deviation of the noise sequence; and replace the original trend estimation values, periodic estimation values, and noise estimation values based on the obtained expected trend estimation values, expected noise estimation values, and expected periodic estimation values; that is, complete the data correction process.

[0009] Furthermore, the process of determining whether the corresponding production sub-process is in a controlled state includes: Construct a blank decision table, map the process control points to the blank nodes in the corresponding process decision table and make directed connections to obtain the corresponding process decision table; Obtain the optimized process monitoring data corresponding to each process control point in the corresponding production sub-process, and input it into a pre-constructed scenario prediction model for production scenario prediction to obtain several process production scenarios corresponding to the corresponding production sub-process; Respectively obtain the parameter safety intervals corresponding to each process control point under the corresponding process production scenario. At the same time, obtain the parameter value intervals corresponding to each process monitoring parameter in the current acquisition period based on the optimized process monitoring data; Obtain the node similarity coefficient between the corresponding production scenario and the process production scenario of the corresponding production sub-process in the current acquisition period based on the parameter value interval and the parameter safety interval; and evaluate whether the corresponding production scenario of the corresponding production sub-process conforms to the process production scenario based on it. If it conforms, mark the corresponding process production scenario as the expected production scenario; Obtain the similarity relationships between all the expected production scenarios, and classify all the expected production scenarios based on them to obtain the corresponding scenario category set; Construct a risk transmission path based on the process decision table, and perform a safety assessment on the corresponding risk transmission path in combination with the scenario category to obtain the comprehensive safety probability corresponding to the corresponding risk transmission path; Based on the process decision table, and combine the comprehensive safety probabilities corresponding to each risk transmission path to obtain the process safety probability corresponding to the corresponding production sub-process; Compare the obtained process safety probability with a pre-constructed safety threshold, and judge whether the corresponding production sub-process is in a controlled state based on the comparison result. If it is not in a controlled state, generate the corresponding risk warning information.

[0010] Furthermore, the process of obtaining the scenario category set includes: Obtain the similarity relationships between different expected production scenarios, and based on them, obtain the similarity classes between the corresponding expected production scenarios; the similarity classes are used to evaluate the similarity degree of the risk characteristics of the same risk transmission path under different corresponding expected production scenarios; if the similarity class is not less than a pre-set similarity class threshold, then divide the corresponding expected production scenarios into the same category, repeat the corresponding process, and obtain the corresponding set of scenario categories.

[0011] Further, the process of performing safety assessment includes: Based on the set of scenario categories, construct the upper approximation risk set and the lower approximation risk set corresponding to each risk transmission path, and substitute the upper approximation risk set and the lower approximation risk set into a pre-defined probability assessment function to obtain the comprehensive safety probability of the corresponding risk transmission path under multiple expected production scenarios.

[0012] Further, the process of obtaining the process safety probability corresponding to the corresponding production sub-process based on the process decision table and combining the comprehensive safety probability corresponding to each risk transmission path includes: Based on the process decision table, obtain the connection mode between the risk transmission paths composed of each process control point, and the connection mode includes parallel connection and series connection; Furthermore, based on the connection mode, solve the comprehensive safety probability corresponding to each risk transmission path in the corresponding production sub-process to obtain the corresponding process safety probability; the solving process is: if the connection mode between the risk transmission paths is series connection, then directly multiply the corresponding comprehensive safety probabilities, and if the connection mode between the risk transmission paths is parallel connection, then add the comprehensive safety probabilities.

[0013] Further, the process of performing process control and feeding back the corresponding process control result includes: When the process warning module receives a risk warning message, perform risk warnings of different degrees based on the warning level in the risk warning message and generate corresponding corrective instructions; The process warning module will synchronously record the running process of the corresponding corrective instructions, and collect the process monitoring data corresponding to the process control points after the adjustment of the corrective instructions based on the data terminal, and feed it back to the process analysis module to determine whether the production sub-process has returned to the controlled state, and repeat the above process until the risk warning message is eliminated.

[0014] The technical effects and advantages of an early warning and control system for milk powder production process of the present invention: Through the whole-process multi-dimensional risk control, the present invention realizes the whole-chain quality and safety monitoring from raw material procurement to finished product packaging; by optimizing the process monitoring data and inputting it into the scenario prediction model, it realizes the accurate identification and early warning of production risks; through the dynamic safety threshold library and fuzzy logic risk assessment, it realizes the intelligent decision-making under complex risk scenarios; and through hierarchical response, it realizes the rapid disposal of production anomalies and the minimization of losses; it realizes the continuous improvement of efficiency and the dynamic optimization of the production process during the production of milk powder. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of an early warning and control system for the production process of milk powder according to the present invention; Figure 2 It is a schematic diagram of an early warning and control method for the production process of milk powder according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment 1 Please refer to Figure 1 As shown, an early warning and control system for the production process of milk powder in this embodiment includes A process acquisition module, which is used to obtain the process control points corresponding to each production sub-process in the whole process of milk powder production, and collect data on the corresponding process control points based on a pre-set data terminal to obtain corresponding process monitoring data; A process analysis module, which is used to divide the process monitoring data into a trend sequence, a cycle sequence and a noise sequence, optimize the corresponding process monitoring data based on them, and judge whether the corresponding production sub-process is in a controlled state based on the optimized process monitoring data. If not, generate corresponding risk warning information; A process warning module, which conducts risk warnings of different levels based on the obtained risk warning information and generates corresponding rectification instructions, and conducts process control on the corresponding production sub-process based on them, and feeds back the corresponding process control results; Each module is connected by wired and / or wireless means to realize data transmission between modules.

[0018] It should be further noted that, in the specific implementation process, the process of obtaining the process monitoring data includes: Obtain the entire process during the production of the target milk powder and divide it into several production sub-processes. The production sub-processes include a raw material procurement sub-process, a raw material storage and transportation sub-process, a raw material processing sub-process, and a finished product packaging and storage sub-process; Obtain the known reasons for milk powder quality risks, conduct a hazard analysis at multiple risk levels for each production sub-process based on them, and obtain the corresponding process control points for each production sub-process based on the hazard analysis results. The process control points refer to the process operations in the corresponding production sub-processes during the milk powder production process that have a direct impact on the quality and safety of the milk powder; The multiple risk levels include chemical, physical, and biological aspects; For example, in the raw material procurement sub-process, the possible hazards of the corresponding milk source include microbial contamination (such as Escherichia coli, Staphylococcus aureus, etc.), veterinary drug residues (such as antibiotics), and heavy metal over-standard (such as lead, mercury, etc.); Then, the raw material inspection process in the corresponding raw material procurement sub-process can be set as a process control point; Set up a data acquisition terminal. The data acquisition terminal collects data from each process control point based on a pre-set acquisition cycle to obtain the corresponding process monitoring data; For example, after setting the raw material acceptance process as a process control point, the corresponding process control parameters can be the microbial index, veterinary drug residue index, and protein content in the milk source during the corresponding raw material acceptance process; In the raw material processing sub-process, the raw material sterilization process can be set as a process control point, and the corresponding process control parameters include sterilization temperature and sterilization time, etc.; Therefore, since there are some differences in the process control parameters corresponding to each process control point in different production sub-processes, the present invention does not specifically limit the process control parameters at the process control points; It should be further noted that, in the specific implementation process, the process of optimizing the corresponding process monitoring data includes: Obtain the process monitoring data corresponding to each process control point and perform data preprocessing on it. The data preprocessing includes sequence synchronization processing and redundancy processing; The sequence synchronization processing refers to mapping the process monitoring data collected at different process control points to the same time series; The redundancy processing refers to removing the duplicate data in the process monitoring data after sequence synchronization; Construct a sliding window with a length of m, where m is a fixed constant; Slide the window based on the sliding window in the process monitoring data after data preprocessing to obtain several monitoring data segments; Obtain the trend coefficient corresponding to each monitoring data segment and use it as the trend estimation value of the data point corresponding to the center point of the corresponding monitoring data segment; The trend estimation value can be used to reflect the overall change trend of the data during the corresponding time period; The formula for obtaining the trend coefficient is: ; In the formula, represents the j-th data point in the monitoring data segment, represents the trend coefficient of the corresponding data point i at the center point within the monitoring data segment; for example, by comparing the trend estimation values at adjacent moments, the overall change trend of the corresponding process monitoring data at the corresponding moment can be obtained; Obtain the trend estimation values corresponding to each data point within the corresponding process monitoring data, and construct a corresponding trend sequence based on them; Perform a difference calculation between the process monitoring data and the corresponding trend sequence to obtain a corresponding detrended sequence; furthermore, divide the detrended sequence into a noise sequence and a periodic sequence based on Fourier transform; the noise sequence is composed of the noise estimation values corresponding to each data point within the corresponding process monitoring data, and the noise estimation value is used to describe the data fluctuation degree of the corresponding data point within the corresponding process monitoring data; the periodic sequence is composed of the periodic estimation values corresponding to each data point within the corresponding process monitoring data, and is used to reflect the periodic change law of the corresponding process monitoring data; (i.e., process monitoring data = trend sequence + noise sequence + periodic sequence); Obtain the noise mean value corresponding to the noise sequence within the corresponding acquisition period and the noise standard deviation , and obtain the noise fluctuation coefficient corresponding to each data point within the corresponding process monitoring data based on them ; where, represents the noise estimation value at the data point corresponding to the center point within the i-th monitoring data segment; Set a fluctuation threshold. If the noise fluctuation coefficient corresponding to the corresponding data point is less than the fluctuation threshold, no other operations are performed. If the noise fluctuation coefficient is not less than the fluctuation threshold, mark the corresponding data point as an abnormal data point; and perform outlier processing on the corresponding abnormal data point. The process of the outlier processing includes: Obtain the trend estimation value, noise estimation value, and periodic estimation value corresponding to the data point at the adjacent moment to the abnormal data point, and combine the linear interpolation algorithm to correct the trend estimation value, noise estimation value, and periodic estimation value at the corresponding abnormal data point; And update the original process monitoring data based on the corrected trend estimation value, noise estimation value, and periodic estimation value; obtain the corresponding optimized process monitoring data; Among them, the process of the data correction includes: Obtain the trend estimation value corresponding to the data point adjacent to the abnormal data point, and combine the linear interpolation algorithm to fit the trend estimation value corresponding to the corresponding abnormal data point to obtain a corresponding expected trend estimation value; at the same time, obtain the periodic change law corresponding to the corresponding periodic sequence, and obtain the corresponding periodic estimation value based on the moment at which the corresponding abnormal data point is located in the corresponding periodic change law; among them, the linear interpolation algorithm is an existing algorithm, and the present invention will not elaborate too much; And randomly generate a noise estimation value according to the noise mean and noise standard deviation of the noise sequence, wherein the randomly generated noise estimation value needs to satisfy the noise mean and noise standard deviation of the corresponding noise sequence; Replace the original trend estimation value, period estimation value, and noise estimation value with the obtained expected trend estimation value, period estimation value, and noise estimation value; that is, complete the data correction process; It should be further noted that in the specific implementation process, the process of determining whether the corresponding generation subprocess is in a controlled state includes: Construct a blank decision table, where several blank nodes are set in the blank decision table, map the obtained process control points to the corresponding blank nodes to obtain the corresponding process nodes; among them, one blank node corresponds to one process control point; Obtain the sequence of each process control point corresponding to the process operations in the corresponding milk powder production process, and connect the corresponding process nodes in a directed manner based on it to obtain the corresponding process decision table; Taking a certain production subprocess as an example, obtain the optimized process monitoring data corresponding to each process control point in the corresponding production subprocess, and input it into a pre-constructed scenario prediction model for production scenario prediction to obtain several process production scenarios corresponding to the corresponding production subprocess; among them, the construction process of the scenario prediction model is prior art, and the present invention will not elaborate too much; the process production scenarios include production scenarios, equipment abnormal production scenarios, process deviation production scenarios, quality risk scenarios, and production efficiency low scenarios; Obtain the set of risk intervals corresponding to each process control point in the corresponding production subprocess under the historical process production scenario; the set of risk intervals refers to the set of continuous or discrete numerical intervals in the range of process control parameter values at the process control points in the corresponding production subprocess that are highly correlated with milk powder quality problems or production abnormalities; Based on it, construct the parameter safety interval corresponding to each process control parameter within the corresponding process control point ; among them, ; in the formula, u represents the index of the process control parameter within the corresponding process node, where u = 1, 2,..., N1, N1 > 0 and N1 is an integer, representing the total number of process control parameters; 、 and 、 respectively represent the left safety boundary value and the right safety boundary value corresponding to the upper approximation interval and the lower approximation interval within the corresponding parameter safety interval; for example: in the raw material inspection stage of the actual milk powder production process, from historical experience and industry standards, it is known that the protein content in the milk source is required to be 2.8% - 3.8%; but in the actual application process of the milk source, high-quality milk sources with a protein content in the range of 3.0% - 3.5% are more preferred; therefore, the parameter safety boundary of the corresponding process control parameter (protein content) can be set as ([3.0%, 3.5%], [2.8%, 3.8%]); Furthermore, based on the optimized process monitoring data, obtain the parameter value intervals corresponding to each process control parameter within each process control point ; where ; for example, in the raw material acceptance stage, the milk source can be sampled at multiple points to obtain the protein content at different sampling points, and the parameter boundary values can be determined based on the maximum and minimum values of the protein content at each sampling point and ; and further determine the parameter boundary values and (that is, the minimum and maximum values of the protein content in the corresponding milk source are and respectively; but through statistical analysis of multiple sampling points, it can be seen that the corresponding protein content data is more concentrated within the range of ) Perform an interval similarity comparison between the corresponding parameter value interval and its corresponding parameter safety interval to obtain the corresponding interval similarity coefficient ; in the formula, represents the decision bias weight; represents the interval similarity coefficient between the parameter value interval and the lower approximation interval within the parameter safety interval; , in the formula, represents the Euclidean distance between the central points within the corresponding lower approximation interval; and respectively represent the interval radii of the lower approximation interval within the parameter value interval and the lower approximation interval within the risk value interval; represents the length of the intersection interval between the lower approximation intervals; where L and U respectively represent the lower approximation interval and the upper approximation interval; for example, the upper approximation interval within the parameter value interval is ; the lower approximation interval is ; where represents the interval similarity coefficient between the parameter value interval and the upper approximation interval within the parameter safety interval; its acquisition process is the same as , and the present invention will not elaborate too much; Furthermore, obtain the interval similarity coefficients corresponding to each process monitoring parameter within the corresponding process control points, and obtain the corresponding node similarity coefficients based on them. ; represents the node similarity coefficient corresponding to the v-th process control point under the corresponding production sub-process, and is used to measure the similarity degree between the process control points in the corresponding process production scenario; Obtain the node similarity coefficients between each process control point in the current production sub-process and the corresponding process control points in the corresponding process production scenario, and comprehensively evaluate whether the current production sub-process conforms to the corresponding process production scenario based on them. If it conforms, no other operations are performed. If it does not conform, mark the corresponding process production scenario as the expected production scenario; among them, the process of comprehensive evaluation includes: Based on the historical milk powder production process, obtain the risk weights corresponding to each process control point, and perform weighted summation on the corresponding node similarity coefficients based on them to obtain the corresponding scenario similarity. If the scenario similarity is not less than a pre-set threshold, it indicates that the value of the corresponding process monitoring parameter is within the parameter safety interval in the corresponding process production scenario, and no other operations are performed; if the scenario similarity is less than the pre-set threshold, it indicates that the value of the corresponding process monitoring parameter is not within the parameter safety interval in the corresponding process production scenario, then it does not conform, and mark the corresponding process production scenario as the expected production scenario; among them, the scenario similarity is used to measure the similarity degree between the production scenario corresponding to the current generation sub-process and the obtained process production scenario; Obtain all the process transfer directions of the corresponding production sub-process based on the process decision table, and construct the corresponding risk transfer path in combination with the corresponding process control points; the risk transfer path refers to the risk propagation direction corresponding to the process control point. For example, when a risk occurs at a certain process control point, it will affect the subsequent process control points according to the process transfer direction, then the propagation path in which the corresponding risk affects the subsequent process control points along the process transfer direction is regarded as the risk transfer path; Construct the corresponding risk assessment set Q = {M, B}; M represents the scenario set composed of expected production scenarios; B represents the risk transfer path set; Arbitrarily select two expected production scenarios, and mark them as x and y respectively; obtain the similarity relationship between the corresponding expected production scenarios based on the risk assessment set ; In the formula, represents the similarity degree between the expected production scenarios x and y under the risk transfer path b; obtained from the node similarities of each process control point in the corresponding risk transfer path; represents the pre-set path similarity threshold; among them, ; In the formula, and respectively represent the node similarity coefficients corresponding to the h-th process control point in the risk transmission path b under the corresponding production sub-processes in the expected production scenarios x and y; h = 1, 2,..., H, where H is a positive integer representing the total number of process control points in the corresponding risk transmission path b; represents the weight coefficient corresponding to the h-th process control point; among them, the closer the distance to the initial process control point in the corresponding risk transmission path, the higher the corresponding weight coefficient; and The acquisition process of can refer to the acquisition process of the above node similarity coefficients ; Obtain the similarity relationship between the corresponding expected production scenarios, and based on it, obtain the similarity classes between the corresponding expected production scenarios ; In the formula, is a pre-set similarity class threshold; represents the number of risk transmission paths; represents that the similarity relationship satisfies The number of scenario pairs of the expected scenarios x and y; The similarity class is used to evaluate the similarity degree of the risk characteristics of the same risk transmission path under the corresponding different expected production scenarios; if the similarity class is not less than the pre-set similarity class threshold, the corresponding expected production scenarios are classified into the same category, and the corresponding process is repeated to obtain the corresponding set of scenario categories; the set of scenario categories consists of several scenario categories, each scenario category contains at least two expected production scenarios, and there is at least one similarity class of risk transmission paths between the corresponding expected production scenarios that is not less than the similarity class threshold; Obtain the upper approximation interval set and the lower approximation interval set of the corresponding risk transmission path under the corresponding set of scenario categories, and the upper approximation interval set and the lower approximation interval set are used to describe the risk boundary corresponding to the corresponding production sub-process; Among them, the expression formula of the upper approximation risk set is: ; The expression formula of the lower approximation risk set is: ; In the formula, X represents the set of scenario categories; Furthermore, based on the risk boundaries described by the upper approximation risk set and the lower approximation risk set, and substituting them into the pre-defined probability evaluation function; obtain the comprehensive safety probability of the corresponding risk transmission path under multiple expected production scenarios The formula for defining the probability evaluation function is: ; In the formula, represents the comprehensive safety probability of the h-th risk transmission path; and represent the contribution value of the expected production scenario g belonging to the lower approximation risk set to the comprehensive safety probability of the risk transmission path h; and represents that the expected production scenario g` belongs to the lower approximation risk set The contribution value to the comprehensive safety probability of the risk transmission path h when; and respectively represent the number of elements in the lower approximation risk set and the upper approximation risk set; Based on the process decision table, obtain the connection methods between the risk transmission paths composed of each process control point, and the connection methods include parallel connection and series connection; Furthermore, based on the connection method, solve the comprehensive safety probability corresponding to each risk transmission path in the corresponding production sub-process to obtain the corresponding process safety probability; the solving process is as follows: if the connection method between the risk transmission paths is series connection, directly multiply the corresponding comprehensive safety probability, if the connection method between the risk transmission paths is parallel connection, add the comprehensive safety probabilities; Set a safety threshold, compare the obtained process safety probability with the corresponding safety threshold, if the process safety probability is not less than the safety threshold, it indicates that the corresponding production sub-process is in a controlled state, then do not perform any other operations; if the process safety probability is less than the safety threshold, it indicates that the corresponding production sub-process is not in a controlled state, then generate corresponding risk warning information based on the process safety probability, and the risk warning information includes mild warning, slightly mild warning, moderate warning, moderately severe warning and severe warning; It should be further noted that in the specific implementation process, the scenario prediction model is pre-constructed based on the process production scenarios corresponding to several historical risk cases of the production sub-process, with the CART decision tree as the basic learner and combined with the random forest algorithm. The corresponding construction process is prior art and will not be elaborated in detail in the present invention.

[0019] It should be further noted that in the specific implementation process, the process of performing process control and feedback of the corresponding process control results includes: When the process warning module receives the risk warning information, perform risk warnings of different degrees based on the warning level in the risk warning information and generate corresponding rectification instructions; An embodiment of the present invention includes: When the risk warning information is a mild warning or a slightly mild warning, feedback the warning information to the operators of the corresponding production sub-process. At the same time, obtain the risk transmission path with the lowest comprehensive safety probability under the corresponding production sub-process, and generate corresponding rectification instructions based on the parameter value range corresponding to the corresponding risk transmission path. For example, in a certain process control point in the raw material processing sub-process, if its corresponding temperature control parameter is close to the lower or upper limit of the parameter value range, then generate a corresponding temperature rectification instruction; When the risk warning information is medium warning, or moderately severe warning, or severe warning, the feedback adjustment module first generates a process deceleration instruction to reduce or stop the production progress of the corresponding production sub-process, and feedbacks the warning information, and simultaneously conducts audible and visual warnings; at the same time, it calls the pre-stored expert knowledge base to conduct a more in-depth analysis of the cause of the risk and gives a more detailed rectification plan; thereby generating corresponding rectification instructions to control the operation process of the corresponding production sub-process; at the same time, the operators of the corresponding process can intervene manually based on the rectification plan; for example: in the raw material storage and operation sub-process, if it is found that the storage humidity of the raw material exceeds the parameter value range, which may lead to the growth of microorganisms, the system will suspend the raw material taking link and, according to the suggestions of the expert knowledge base, guide the operators to take measures such as ventilation and dehumidification or adjusting the parameters of the storage equipment to solve the corresponding problems.

[0020] Furthermore, the process warning module will synchronously record the operation process of the corresponding rectification instructions, and collect the process monitoring data corresponding to the process control points after the rectification instructions are adjusted based on the data terminal, and feedback it to the process analysis module to determine whether the production sub-process has returned to the controlled state, and repeat the above process until the risk warning information is eliminated.

[0021] The present invention effectively avoids the occurrence of quality problems, improves production efficiency and product quality by real-time monitoring and analyzing data in the milk powder production process, taking corresponding rectification measures in a timely manner, and simultaneously realizes the intelligent management of the milk powder production process, reduces the dependence on manual intervention, and improves the accuracy and efficiency of management.

[0022] Embodiment 2 Please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, please refer to the description content of Embodiment 1. A warning and control method for the milk powder production process is provided, including Step 1: Obtain the process control points corresponding to each production sub-process in the whole milk powder production process, and collect data for the corresponding process control points based on the pre-set data terminal to obtain the corresponding process monitoring data; Step 2: Divide the process monitoring data into a trend sequence, a periodic sequence, and a noise sequence, and optimize the corresponding process monitoring data based on them. Based on the optimized process monitoring data, determine whether the corresponding production sub-process is in a controlled state. If not, generate corresponding risk warning information; Step 3: Conduct risk warnings of different degrees based on the obtained risk warning information and generate corresponding rectification instructions, and conduct process control on the corresponding production sub-process based on them, and feedback the corresponding process control results.

[0023] Embodiment 3 This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the above-provided warning and control system for the milk powder production process.

[0024] Since the electronic device introduced in this embodiment is the electronic device used to implement a warning and control system for the milk powder production process in the embodiments of the present application, based on the warning and control system for the milk powder production process introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiments of the present application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the warning and control system for the milk powder production process in the embodiments of the present application, it falls within the scope of protection of the present application.

[0025] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain a formula that is closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0026] The above are only the preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for ordinary technical users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. An early warning and control system for the milk powder production process, characterized in that, including a process acquisition module, configured to obtain process control points corresponding to each production subprocess in the entire process of milk powder production, and based on a preset data terminal, perform data acquisition on the corresponding process control points to obtain corresponding process monitoring data; a process analysis module, configured to divide the process monitoring data into a trend sequence, a periodic sequence, and a noise sequence, and based on these, perform data optimization on the corresponding process monitoring data, and based on the process monitoring data after data optimization, determine whether the corresponding production subprocess is in a controlled state. If not, generate corresponding risk warning information; a process warning module, configured to perform risk warnings at different levels based on the obtained risk warning information and generate corresponding rectification instructions, and based on these, perform process control on the corresponding production subprocess and provide feedback on the corresponding process control results.

2. The early warning and control system for the milk powder production process according to claim 1, characterized in that, The process of obtaining the process monitoring data includes: obtain the entire process in the production process of the target milk powder and divide it into several production subprocesses; obtain known reasons for milk powder quality risks, and based on these, perform hazard analysis at multiple risk levels on each production subprocess to obtain process operations that have a direct impact on the corresponding production subprocess in the milk powder production process; and set them as process control points; set up a data acquisition terminal, and the data acquisition terminal performs data acquisition on each process control point based on a preset acquisition period to obtain corresponding process monitoring data.

3. The early warning and control system for the milk powder production process according to claim 2, characterized in that, The process of performing data optimization on the corresponding process monitoring data includes: obtain the process monitoring data corresponding to each process control point and perform data preprocessing on it. The data preprocessing includes sequence synchronization processing and redundancy processing; divide the process monitoring data after data preprocessing into a trend sequence, a periodic sequence, and a noise sequence; obtain the noise mean and noise standard deviation corresponding to the corresponding noise sequence, and based on these, obtain the noise fluctuation coefficient corresponding to each data point in the corresponding process monitoring data; and based on the noise fluctuation coefficient, determine whether the corresponding data point is an abnormal data point. If so, perform outlier processing on the corresponding abnormal data point; obtain corresponding optimized process monitoring data.

4. A warning and control system for the milk powder production process according to claim 3, characterized in that, The trend sequence is composed of trend estimated values corresponding to each data point; the noise sequence is composed of noise estimated values corresponding to each data point in the corresponding process monitoring data; the periodic sequence is composed of periodic estimated values corresponding to each data point in the corresponding process monitoring data.

5. The early warning and control system for the milk powder production process according to claim 4, wherein The process of outlier processing includes: obtain the trend estimated value, noise estimated value, and periodic estimated value corresponding to the data point at the adjacent moment to the abnormal data point, and based on these, perform data correction on the trend estimated value, noise estimated value, and periodic estimated value at the corresponding abnormal data point; and based on the corrected trend estimated value, noise estimated value, and periodic estimated value, perform data update on the original process monitoring data; obtain corresponding optimized process monitoring data; The process of data correction includes: Obtain the trend estimation values corresponding to the data points adjacent to the abnormal data points, and combine with the linear interpolation algorithm to perform data fitting on the trend estimation values corresponding to the corresponding abnormal data points to obtain the corresponding expected trend estimation values; at the same time, obtain the periodic change rules corresponding to the corresponding periodic sequences, and obtain the corresponding expected periodic estimation values based on the moments at which the corresponding abnormal data points are located in the corresponding periodic change rules; randomly generate an expected noise estimation value according to the mean value and noise standard deviation of the noise sequence; and replace the original trend estimation value, periodic estimation value, and noise estimation value based on the obtained expected trend estimation value, expected noise estimation value, and expected periodic estimation value; that is, complete the data correction process.

6. The early warning and control system for the milk powder production process according to claim 5, wherein, The process of determining whether the corresponding production sub-process is in a controlled state includes: Construct a blank decision table, map the process control points to the blank nodes in the corresponding process decision table and make directed connections to obtain the corresponding process decision table; Obtain the optimized process monitoring data corresponding to each process control point in the corresponding production sub-process, and input it into the pre-constructed scenario prediction model for production scenario prediction to obtain several process production scenarios corresponding to the corresponding production sub-process; Respectively obtain the parameter safety intervals corresponding to each process control point under the corresponding process production scenario. At the same time, based on the optimized process monitoring data, obtain the parameter value intervals corresponding to each process monitoring parameter in the current collection period; Obtain the node similarity coefficient between the corresponding production scenario of the corresponding production sub-process and the process production scenario under the current collection period based on the parameter value interval and the parameter safety interval; and evaluate whether the corresponding production scenario of the corresponding production sub-process conforms to the process production scenario based on it. If it conforms, mark the corresponding process production scenario as the expected production scenario; Obtain the similarity relationships between the various expected production scenarios, and classify all the expected production scenarios based on them to obtain the corresponding set of scenario categories; Construct a risk transmission path based on the process decision table, and perform a safety assessment on the corresponding risk transmission path in combination with the scenario category to obtain the comprehensive safety probability corresponding to the corresponding risk transmission path; Based on the process decision table, and combine the comprehensive safety probabilities corresponding to each risk transmission path to obtain the process safety probability corresponding to the corresponding production sub-process; Compare the obtained process safety probability with the pre-constructed safety threshold, and judge whether the corresponding production sub-process is in a controlled state based on the comparison result. If it is not in a controlled state, generate the corresponding risk warning information.

7. The warning and control system for the milk powder production process according to claim 6, characterized in that, The process of obtaining the set of scenario categories includes: Obtain the similarity relationships between different expected production scenarios, and obtain the similarity classes between the corresponding expected production scenarios based on them; the similarity classes are used to evaluate the similarity degree of the risk characteristics of the same risk transmission path under the corresponding different expected production scenarios; if the similarity class is not less than the pre-set similarity class threshold, divide the corresponding expected production scenarios into the same category, and repeat the corresponding process to obtain the corresponding set of scenario categories.

8. The early warning and control system for the milk powder production process according to claim 7, characterized in that The process of performing a safety assessment includes: Construct the upper approximation risk set and the lower approximation risk set corresponding to each risk transmission path based on the set of scenario categories, and substitute the upper approximation risk set and the lower approximation risk set into a pre-defined probability evaluation function to obtain the comprehensive safety probability of the corresponding risk transmission path under multiple expected production scenarios.

9. The warning and control system for the milk powder production process according to claim 8, wherein, The process of obtaining the process safety probability corresponding to the corresponding production sub-process based on the process decision table and combining the comprehensive safety probabilities corresponding to each risk transmission path includes: Based on the process decision table, obtain the connection modes between the risk transmission paths composed of each process control point, and the connection modes include parallel connection and series connection; Furthermore, based on the connection mode, solve the comprehensive safety probabilities corresponding to each risk transmission path in the corresponding production sub-process to obtain the corresponding process safety probability; the solving process is as follows: if the connection mode between the risk transmission paths is series connection, directly multiply the corresponding comprehensive safety probabilities, and if the connection mode between the risk transmission paths is parallel connection, add the comprehensive safety probabilities.

10. The warning and control system for the milk powder production process according to claim 9, characterized in that, The process of performing process control and feedback of the corresponding process control results includes: When the process warning module receives risk warning information, perform risk warnings of different degrees based on the warning level in the risk warning information and generate corresponding corrective instructions; The process warning module will synchronously record the operation process of the corresponding corrective instructions, and collect the process monitoring data corresponding to the process control points after the corrective instructions are adjusted based on the data terminal, and feedback it to the process analysis module to determine whether the production sub-process has returned to the controlled state, and repeat the above process until the risk warning information is eliminated.

Images