Audit tracking method and system for medical pure water machine

Abstract

The application discloses a kind of medical pure water machine audit tracking methods, more specifically related to audit tracking technical field, including when identifying that medical pure water machine exists abnormal parameter, call the user of last period to current detection period to abnormal parameter is modified, by the time point clustering analysis of modification user to pure water machine modification, and by comparing the similarity of operation characteristics between modification user, determine the historical operation information of modification user, according to the time series analysis of modification user modification abnormal parameter, determine the historical habit information of modification user, the historical operation information and historical habit information of modification user are comprehensively analyzed, determine suspect user, extract the correlation information and importance information of suspect user each time to medical pure water machine modification, determine the potential abnormal modification record of suspect user, the application helps to improve the operation safety of medical pure water machine, potential abnormal operation is promptly discovered, improve the efficiency of audit tracking.

Description

Technical Field

[0001] The present invention relates to the field of audit tracking technology, and more particularly, to an audit tracking method and system for a medical pure water machine. Background Art

[0002] In order to ensure the quality of the water used in medical pure water machines, the equipment parameters of medical pure water machines need to be adjusted and maintained regularly. However, as the frequency of use increases, various parameter abnormalities may occur in medical pure water machines, leading to water quality problems. This not only affects the normal operation of the equipment, but may also affect the safety of patients. In actual operation, the maintenance and operation records of medical pure water machines are usually managed and operated by multiple staff members. However, due to the high mobility of staff and the complexity of equipment operation, some operating users may not be able to detect and correct abnormal adjustments to the equipment in time, thereby affecting the reliability of the equipment and water quality. In addition, there is often a lack of in-depth analysis of the modification records of medical pure water machine parameters, making it difficult to identify potential abnormal modification behaviors in the operation history.

[0003] In order to solve the above-mentioned defects, a technical solution is now provided. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a medical water purifier audit tracking method and system to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A medical water purifier audit tracking method specifically includes the following steps:

[0007] S1: Periodically identify abnormal parameters of the medical water purifier. When abnormal parameters of the medical water purifier are identified, retrieve the users who modified the abnormal parameters from the previous cycle to the current detection cycle;

[0008] S2: By clustering the time points at which the user modified the water purifier, and comparing the similarity of the operating characteristics between the users, the user's historical operation information is determined. Based on the time series analysis of the abnormal parameters modified by the user, the user's historical habit information is determined.

[0009] S3: Identify the suspected user through comprehensive analysis of the user's historical operation information and historical habit information, and extract the relevant information and importance information of each modification made by the suspected user to the medical pure water machine;

[0010] S4: Determine the potential abnormal modification records of the suspicious user through the associated information and importance information of the suspicious user's modifications to the medical pure water machine.

[0011] In a preferred embodiment, periodically identifying abnormal parameters of a medical water purifier includes:

[0012] By using a card reader in combination with read-write cards of different authority levels to set equipment parameters, and by periodically conducting quality inspections on the medical pure water machine, the safety of the medical pure water machine is ensured. When problems arise in the quality inspection of the medical pure water machine, further inspections are conducted to determine the parameter settings of the medical pure water machine, and the abnormal parameter settings of the medical pure water machine are obtained. Based on the abnormal parameter settings of the medical pure water machine, users who have modified the abnormal parameters from the previous cycle to the current inspection cycle are retrieved.

[0013] In a preferred embodiment, determining to modify the historical operation information of the user includes:

[0014] Modify the user's historical operation information through the cluster center time distance difference coefficient and operation feature similarity coefficient;

[0015] The logic for obtaining the time-distance difference coefficient of the cluster center is as follows: obtaining the historical operation records of users who modified abnormal parameters from the previous cycle to the current detection cycle, marking the users who modified abnormal parameters from the previous cycle to the current detection cycle as modifying users, determining the time points when each modifying user modified the abnormal parameters of the medical pure water machine parameters, using the DBSCAN clustering algorithm to divide the operation time points of different modifying users, and setting the neighborhood radius to define the maximum distance between time points and the minimum number of points to determine the minimum number of time points required for a cluster. Areas with less than the minimum number of points are regarded as noise;

[0016] Perform DBSCAN clustering on the modification time points of each user in a day to obtain clusters of abnormal parameters of the medical water purifier. Each cluster represents a regular time period of the modification user's operation, and the noise represents some abnormal operation time points.

[0017] The cluster center of the cluster is approximated by calculating the mean of all time points in the cluster, and the cluster center of each cluster is marked as: SJ n , where n is the number of clusters after DBSCAN clustering;

[0018] According to the operation time point of the modified user, determine whether there is a noise point in the modified user, and mark the noise point of the modified user as: Where i = 1, 2, 3, ..., I, where I is a positive integer and I is the number of the user who modified the abnormal parameters from the previous cycle to the current detection cycle; m = 1, 2, 3, ..., M, where M is a positive integer and m is the number of the noise point where the i-th modifying user exists;

[0019] According to the noise point of the modified user, the minimum time difference between the noise point of the modified user and the cluster center of each cluster is determined, and the minimum time difference between the noise point of the modified user and the cluster center of each cluster is marked as: in, If there is no noise point for the modified user, the minimum time difference between the noise point of the modified user and the cluster center of each cluster is set to 0;

[0020] Eliminate the noise points of user modification and mark the user modification time points according to the user modification time points: Where j = 1, 2, 3, ..., J, where J is a positive integer and j is the number of the i-th operation time point of the user modification. The minimum time difference between the operation time point of the user modification and the cluster center of each cluster is determined, and the minimum time difference between the operation time point of the user modification and the cluster center of each cluster is marked as: in,

[0021] Calculate the cluster center time distance difference coefficient, the calculation formula is: Among them, cy i is the cluster center time distance difference coefficient of the i-th modified user.

[0022] In a preferred embodiment, the logic for obtaining the operation feature similarity coefficient is:

[0023] Determine the operation characteristics of the modifying user. The operation characteristics of the modifying user include modification frequency, modification amplitude, and operation duration. The modification frequency is determined by determining the number of times the abnormal parameters are modified from the previous cycle to the current detection cycle. The modification amplitude is determined by recording the change amplitude of each abnormal parameter modification by the modifying user and determining the maximum amplitude of the abnormal parameter modification by the modifying user. The maximum amplitude of the abnormal parameter modification by the modifying user is used as the modification amplitude in the operation characteristics. The operation duration is determined by recording the average duration of each modification by the modifying user. The operation characteristics of the i-th modifying user are represented by a vector: X i =(pl i ,fd i ,sc i );

[0024] The similarity between the modifying user and the operation feature vectors of other modifying users is calculated by cosine similarity. The calculation formula is: Among them, XS ik is the similarity between the operation features of the i-th modifying user and the k-th other modifying user, X k is the operation feature vector of other modifying users except the i-th modifying user, k=1, 2, 3, ..., K, K is a positive integer, and k is the number of other modifying users except the i-th modifying user;

[0025] Calculate the operation feature similarity coefficient, the calculation formula is: Among them, TZXS i is the operation feature similarity coefficient.

[0026] In a preferred embodiment, determining to modify the user's historical habit information includes:

[0027] The logic for obtaining the stability coefficient of the operation characteristic is as follows: according to the operation of modifying the abnormal parameters of the medical pure water machine by the user in the previous cycle, the difference of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle is obtained, and the difference of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle is marked as: CZ r , where r = 1, 2, 3, ..., R, R is a positive integer, and r is the number of each time the user modified the medical pure water machine in the previous cycle;

[0028] According to the difference in the value of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle, the time series of the difference in the value of the modification of the abnormal parameters of the user in the previous cycle is determined, and the ARMA model is trained based on the time series data to determine the orders P and Q of the ARMA model. The difference in the value of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle is used to predict the difference in the value of each modification of the abnormal parameters of the medical pure water machine by the user from the previous cycle to the current cycle;

[0029] The ARMA model expression obtained by training is: Among them, α is a constant term, which represents the offset of the time series, p = 1, 2, 3, ..., P, p is the number of the order P of the ARMA model, q = 1, 2, 3, ..., Q, q is the number of the order Q of the ARMA model, φ p is the coefficient of the MA part in the ARMA model, which represents the relationship with the residuals of the past p moments, θ q is the coefficient of the AR part in the ARMA model, which represents the relationship with the value of the past q moments, ε r is the white noise term;

[0030] The ARMA model obtained through training predicts the difference between the user's modification of the abnormal parameters of the medical pure water machine from the previous cycle to the current cycle. The expression is: Among them, d R+t To predict the difference in the user's t-th modification of the abnormal parameters of the medical pure water machine, d R+t-p is the known observation value of the time series data, ε R+t-q is the prediction error of time series data, t = 1, 2, 3, ..., T, T is a positive integer, and t is the number of modified differences of the required prediction anomaly parameters;

[0031] Obtain the difference in the actual modification of the abnormal parameters of the medical pure water machine by the modifying user from the previous cycle to the current detection cycle, and mark the difference in the actual modification of the abnormal parameters of the medical pure water machine by the modifying user from the previous cycle to the current detection cycle as: DQ t ;

[0032] By calculating the variance of the residual error of the modified user's abnormal parameter difference, the operating characteristic stability coefficient is determined, and the operating characteristic stability coefficient of the i-th modified user is marked as: WD i ,in,

[0033] In a preferred embodiment, determining a suspicious user includes:

[0034] By modifying the user's historical operation information and historical habit information through comprehensive analysis, the cluster center time distance difference coefficient, operation feature similarity coefficient and operation feature stability coefficient are weighted and calculated to build a user evaluation model and generate a user evaluation coefficient. The calculation formula of the user evaluation coefficient is: Among them, pg i is the user evaluation coefficient of the i-th modifying user, β1, β2, and β3 are the proportional coefficients of the cluster center time distance difference coefficient, the operation feature similarity coefficient, and the operation feature stability coefficient, respectively. β1, β2, and β3 are all greater than 0;

[0035] Set a user evaluation coefficient threshold, compare the user evaluation coefficients of all modified users with the user evaluation coefficient threshold, and mark the modified users whose user evaluation coefficients are greater than the user evaluation coefficient threshold as suspicious users. If the user evaluation coefficient of the modified user is less than the user evaluation coefficient threshold, no mark will be made.

[0036] In a preferred embodiment, the associated information of each modification made by the suspicious user to the medical pure water machine includes:

[0037] The associated information of the suspected user's modification of the medical pure water machine is represented by the correlation closeness coefficient;

[0038] The logic for obtaining the correlation closeness coefficient is as follows: constructing a knowledge graph of medical water purifier parameters, wherein the basic unit in the knowledge graph of medical water purifier parameters is a node, a node represents an entity, and represents the parameters of the medical water purifier. Nodes are connected by edges, which represent the relationship between the medical water purifier parameters. The edges between nodes have a weight value, which represents the strength of the relationship between the nodes.

[0039] Obtain the parameters of the medical pure water machine modified by the suspicious user each time, and represent the parameters of the medical pure water machine modified by the suspicious user each time as a set: XG g={GB1, GB2, GB3,..., GB H}, where g = 1, 2, 3, ..., G, G is a positive integer, g is the number of times the suspected user has modified the medical pure water machine from the previous cycle to the current detection cycle, GB1, GB2, GB3, ..., GB H For each parameter modified by the suspicious user, the distance between the abnormal parameters of the medical pure water machine and the parameters modified by the suspicious user is determined through the shortest path analysis in the graph algorithm, and the distance between the abnormal parameters of the medical pure water machine and the parameters modified by the suspicious user is marked as: JL h ;

[0040] Calculate the correlation closeness coefficient, the calculation formula is: Among them, JM g The correlation closeness coefficient of the medical pure water machine modified by the suspicious user for the gth time, h = 1, 2, 3, ..., H, H is a positive integer, and h is the number of the medical pure water machine parameter modified for the gth time.

[0041] In a preferred embodiment, the importance information of each modification made by the suspected user to the medical water purifier includes:

[0042] The importance of the modification of the medical water purifier by the doubtful user is expressed by the risk important regression coefficient;

[0043] The logic for obtaining the risk important regression coefficient is as follows: determining the parameters of the medical water purifier, collecting the historical data of the parameters of the medical water purifier, and obtaining the water quality data corresponding to the parameters of each medical water purifier, constructing a logistic risk important regression model through a logistic regression model, generating a risk important regression coefficient, and quantifying the importance of each parameter through the risk important regression coefficient;

[0044] The expression of the Logistic risk important regression model is: Among them, HG is the risk important regression coefficient of the Logistic risk important regression model, CS1, CS2, CS3, ..., CS B are the values ​​of all parameters of the medical water purifier, λ1, λ2, λ3, ..., λ B is the regression coefficient, which indicates the influence of each medical water purifier parameter on water quality risk;

[0045] Obtain the parameters of the medical pure water machine after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle, and bring the parameters of the medical pure water machine after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle into the Logistic risk important regression model to obtain the risk important regression coefficient after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle, and mark the risk important regression coefficient after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle as: HG g .

[0046] In a preferred embodiment, determining the potential abnormal modification records of the suspicious user includes:

[0047] Through a comprehensive analysis of the relevance and importance information of the suspicious users' modifications to the medical pure water machine, the correlation closeness coefficient and the risk important regression coefficient are weighted and calculated to construct a modification evaluation model and generate a modification evaluation coefficient. The calculation formula of the modification evaluation coefficient is: Among them, pg g is the modification evaluation coefficient of the g-th modification by the suspicious user, γ1 and γ2 are the proportional coefficients of the correlation closeness coefficient and the risk important regression coefficient, respectively, and γ1 and γ2 are both greater than 0;

[0048] Set a modification evaluation coefficient threshold and compare the modification evaluation coefficient of each modification made by the suspicious user from the previous cycle to the current detection cycle with the modification evaluation coefficient threshold. If the modification evaluation coefficient is greater than the modification evaluation coefficient threshold, the modification is marked; if the modification evaluation coefficient is less than the modification evaluation coefficient threshold, the modification is not marked.

[0049] In a preferred embodiment, a medical water purifier audit tracking system includes an anomaly identification module, a data acquisition module, a user evaluation module, and a record audit evaluation module, with signal connections between the modules;

[0050] The abnormality identification module is used to periodically monitor the operating status of the medical water purifier and identify abnormal parameters that occur in the current cycle;

[0051] The data collection module is used to collect and modify the user's historical operation information and historical habit information, and collect the associated information and importance information of the medical pure water machine modification;

[0052] The user evaluation module is used to build a user evaluation model, generate a user evaluation coefficient, and compare the user evaluation coefficient with the user evaluation coefficient threshold to identify suspicious users;

[0053] The record audit assessment module is used to build a modification assessment model, generate a modification assessment coefficient, and compare the modification assessment coefficient of the suspicious user with the modification assessment coefficient threshold to determine the potential abnormal modification records of the suspicious user.

[0054] The technical effects and advantages of the present invention are as follows:

[0055] The present invention periodically identifies abnormal parameters of medical water purifiers, conducts in-depth analysis based on the user's modification behavior, and then identifies potential abnormal modification records and ultimately determines the suspicious user, which helps to improve the operational safety of medical water purifiers, timely discover potential abnormal operations, and improve the efficiency of audit tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0057] Figure 1 A schematic flow chart of a method for auditing and tracking a medical water purifier according to the present invention;

[0058] Figure 2 The figure is a schematic structural diagram of an audit tracking system for a medical water purifier according to the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] Example 1

[0061] like Figure 1 A schematic flow chart of a method for auditing and tracking a medical water purifier according to the present invention is provided, which specifically includes the following steps:

[0062] S1: Periodically identify abnormal parameters of the medical water purifier. When abnormal parameters of the medical water purifier are identified, retrieve the users who modified the abnormal parameters from the previous cycle to the current detection cycle;

[0063] S2: By clustering the time points at which the user modified the water purifier, and comparing the similarity of the operating characteristics between the users, the user's historical operation information is determined. Based on the time series analysis of the abnormal parameters modified by the user, the user's historical habit information is determined.

[0064] S3: Determine the suspicious user by comprehensive analysis of the user's historical operation information and historical habit information, extract the associated information and importance information of the suspicious user's each modification of the medical pure water machine;

[0065] S4: Determine the potential abnormal modification record of the suspicious user through the associated information and importance information of the suspicious user's modification of the medical pure water machine.

[0066] In order to ensure the safety of water quality, the compliance of equipment and the traceability of operation, and to protect the health and safety of medical environment, the use of card reader combined with different permission level read-write cards for equipment parameter setting can improve the safety of medical pure water machine, which includes:

[0067] Different permission users use different read-write cards, and different read-write cards have the modification of all parameters and system settings under this permission, and limit the parameters and system settings outside the permission;

[0068] When modifying the parameters of the medical pure water machine, the user is authenticated, the card reader reads the card information of the permission card and decides all parameters and system settings of the user under the permission according to the user's permission level;

[0069] After each card swiping operation, the system will record the timestamp, user ID, permission level, operation type (reading, modification, etc.) of the relevant operation, so as to query and audit later, and generate the historical operation record of the medical pure water machine.

[0070] Through periodic quality detection of the medical pure water machine, the safety of the medical pure water machine is ensured, and when the quality detection of the medical pure water machine appears problems, the parameter setting of the medical pure water machine is further detected to obtain the abnormal parameter setting of the medical pure water machine, wherein the content of the quality detection of the medical pure water machine includes:

[0071] Water quality detection: including water purity (such as conductivity, TDS, etc.), content of microorganisms such as bacteria and viruses, pH value, etc.;

[0072] Equipment performance detection: whether the running pressure, flow, temperature, etc. of the equipment meet the design requirements;

[0073] Equipment health state detection: including whether the filter core pollution degree and membrane aging meet the expected service life.

[0074] The detection of the parameter setting of the medical pure water machine includes water flow, pressure, temperature, pH value, conductivity, etc. which need to be monitored in real time, and threshold value is set to judge whether there is abnormality.

[0075] It should be noted that periodic quality inspections focus on the overall operating status of the equipment and water quality indicators, while parameter abnormalities may not directly lead to quality inspection abnormalities. However, the occurrence of quality inspection abnormalities must be related to certain parameter abnormalities. Therefore, the quality inspection results can be used as a trigger point to further trigger in-depth analysis of parameter abnormalities.

[0076] According to the abnormal parameter settings of the medical pure water machine, the users who modified the abnormal parameters from the previous cycle to the current detection cycle are retrieved, and the operation characteristics of the modified users in the historical operation records of the medical pure water machine are obtained. The historical operation records of the modified users are collected, and based on the usage habits of the modified users, the historical habit information of the modified users is collected. The historical operation information of the modified users is expressed by the cluster center time distance difference coefficient and the operation feature similarity coefficient, and the historical habit information of the modified users is expressed by the operation feature stability coefficient.

[0077] The logic for obtaining the time-distance difference coefficient of the cluster center is as follows: obtaining the historical operation records of users who modified abnormal parameters from the previous cycle to the current detection cycle, marking the users who modified abnormal parameters from the previous cycle to the current detection cycle as modifying users, determining the time points when each modifying user modified the abnormal parameters of the medical pure water machine parameters, using the DBSCAN clustering algorithm to divide the operation time points of different modifying users, and setting the neighborhood radius to define the maximum distance between time points and the minimum number of points to determine the minimum number of time points required for a cluster. Areas with less than the minimum number of points are regarded as noise;

[0078] It should be noted that the modification of medical water purifier parameters is usually regular, and the parameters of the medical water purifier need to be modified or checked during a specific time period of the day. Therefore, the clustering algorithm can be used to identify users who modify or check parameters more frequently within a day.

[0079] Perform DBSCAN clustering on the modification time points of each user in a day to obtain clusters of abnormal parameters of the medical water purifier. Each cluster represents a regular time period of the modification user's operation, and the noise represents some abnormal operation time points.

[0080] The cluster center of the cluster is approximated by calculating the mean of all time points in the cluster, and the cluster center of each cluster is marked as: SJ n , where n is the number of clusters after DBSCAN clustering;

[0081] According to the operation time point of the modified user, determine whether there is a noise point in the modified user, and mark the noise point of the modified user as: Where i = 1, 2, 3, ..., I, where I is a positive integer and I is the number of the user who modified the abnormal parameters from the previous cycle to the current detection cycle; m = 1, 2, 3, ..., M, where M is a positive integer and m is the number of the noise point where the i-th modifying user exists;

[0082] According to the noise point of the modified user, the minimum time difference between the noise point of the modified user and the cluster center of each cluster is determined, and the minimum time difference between the noise point of the modified user and the cluster center of each cluster is marked as: in, If there is no noise point for the modified user, the minimum time difference between the noise point of the modified user and the cluster center of each cluster is set to 0;

[0083] Eliminate the noise points of user modification and mark the user modification time points according to the user modification time points: Where j = 1, 2, 3, ..., J, where J is a positive integer and j is the number of the i-th operation time point of the user modification. The minimum time difference between the operation time point of the user modification and the cluster center of each cluster is determined, and the minimum time difference between the operation time point of the user modification and the cluster center of each cluster is marked as: in,

[0084] Calculate the cluster center time distance difference coefficient, the calculation formula is: Among them, cy i is the cluster center time distance difference coefficient of the i-th modified user.

[0085] From the formula, we can see that the larger the time-distance difference coefficient of the cluster center of the modified user, the more irregular the time of the modified user's operation of the water purifier may be, which means that the operations performed by the user at different time points are significantly different from the operation patterns of most users, thereby increasing the possibility of the user causing equipment parameter abnormalities.

[0086] The logic for obtaining the similarity coefficient of the operation feature is as follows: determining the operation feature of the modifying user, the operation feature of the modifying user includes modification frequency, modification amplitude, and operation duration. The modification frequency is determined by determining the number of times the abnormal parameters are modified from the previous cycle to the current detection cycle. The modification amplitude is determined by recording the change amplitude of each abnormal parameter modification by the modifying user, determining the maximum amplitude of the abnormal parameter modification by the modifying user, and taking the maximum amplitude of the abnormal parameter modification by the modifying user as the modification amplitude in the operation feature. The operation duration is determined by recording the average duration of each modification by the modifying user. The operation feature of the i-th modifying user is represented by a vector as: X i =(pl i ,fd i ,sc i );

[0087] The similarity of the operation feature vectors of the user and other modification users is calculated by a cosine similarity calculation, and the calculation formula is: Wherein, Xs ik is the similarity of the operation feature of the i th modification user and the k th other modification user, X k is the operation feature vector of the other modification user except the i th modification user, k = 1, 2, 3, …, K, K is a positive integer, and k is the number of the other modification user except the i th modification user;

[0088] The operation feature similarity coefficient is calculated, and the calculation formula is: Wherein, TZXS i is the operation feature similarity coefficient.

[0089] As can be seen from the formula, the greater the operation feature similarity coefficient, the more similar the operation behavior of the modification user for the medical water machine from the last cycle to the current cycle to the operation behavior of the same user, indicating that the modification user may operate according to the system expected regular operation process, the operation habit is relatively stable and regular, which means that the operation of the user does not deviate from the normal behavior, and conforms to the normal behavior.

[0090] The operation feature stability coefficient is obtained according to the operation of the modification user on the abnormal parameters of the medical water machine in the last cycle, and the difference value of the modification user on the abnormal parameters of the medical water machine in the last cycle is obtained. The difference value of the modification user on the abnormal parameters of the medical water machine in the last cycle is marked as: CZ r , wherein r = 1, 2, 3, …, R, R is a positive integer, and r is the number of the modification user modifying the medical water machine in the last cycle;

[0091] According to the difference value of the modification user on the abnormal parameters of the medical water machine in the last cycle, the time sequence of the difference value of the modification user on the abnormal parameters of the medical water machine in the last cycle is determined, and the ARMA model is trained according to the time sequence data to determine the order P and Q of the ARMA model. The difference value of the modification user on the abnormal parameters of the medical water machine in the last cycle is predicted by the difference value of the modification user on the abnormal parameters of the medical water machine in the last cycle.

[0092] It should be noted that the order p and q of the ARMA model are set by professional staff in the field, and are usually determined by ACF (autocorrelation function) and PACF (partial autocorrelation function) graphs combined with experience.

[0093] The expression of the trained ARMA model is: wherein, a is a constant term, representing the offset of the time series, p = 1, 2, 3, …, P, p is the number of the order P of the ARMA model, q = 1, 2, 3, …, Q, q is the number of the order Q of the ARMA model, φ p is the coefficient of the MA part of the ARMA model, representing the relationship with the residual of the past p time, θ q is the coefficient of the AR part of the ARMA model, representing the relationship with the value of the past q time, ε r is a white noise term;

[0094] The ARMA model obtained by training is used to predict the difference of the modification of the abnormal parameters of the medical water machine by the modification user in the last period to the current period, and the expression is: wherein, d R+t is the predicted difference of the modification of the abnormal parameters of the medical water machine by the modification user for the tth time, d R+t-p is the known observation value of the time series data, ε R+t-q is the prediction error of the time series data, t = 1, 2, 3, …, T, T is a positive integer, t is the number of the required prediction of the abnormal parameter modification difference, and the number of the required prediction of the abnormal parameter modification difference is usually the same as the number of the actual modification of the abnormal parameters of the medical water machine by the modification user in the last period to the current detection period;

[0095] The actual difference of the modification of the abnormal parameters of the medical water machine by the modification user in the last period to the current detection period is obtained, and the actual difference of the modification of the abnormal parameters of the medical water machine by the modification user in the last period to the current detection period is marked as: DQ t ;

[0096] The operation characteristic stability coefficient is determined by calculating the variance of the residual of the abnormal parameter difference of the modification user, and the operation characteristic stability coefficient of the ith modification user is marked as: WD i , wherein,

[0097] As can be seen from the formula, the greater the operation characteristic stability coefficient of the modification user, the greater the prediction error of the ARMA model, indicating that the modification habits and rules of the modification user may have changed.

[0098] Through comprehensive analysis of the historical operation information and historical habit information of the modification user, the cluster center time distance difference coefficient, the operation characteristic similarity coefficient and the operation characteristic stability coefficient are weighted calculated, the user evaluation model is constructed, and the user evaluation coefficient is generated. The calculation formula of the user evaluation coefficient is: wherein, pg iis the user evaluation coefficient of the i-th modifying user, β1, β2, and β3 are the proportional coefficients of the cluster center time distance difference coefficient, the operation feature similarity coefficient, and the operation feature stability coefficient, respectively. β1, β2, and β3 are all greater than 0.

[0099] Set a user evaluation coefficient threshold, compare the user evaluation coefficients of all modified users with the user evaluation coefficient threshold, and mark the modified users whose user evaluation coefficients are greater than the user evaluation coefficient threshold as suspicious users, indicating that the modified users are more likely to violate regulations on the medical pure water machine. If the user evaluation coefficient of the modified user is less than the user evaluation coefficient threshold, no mark will be made.

[0100] Based on the abnormal parameters of the medical pure water machine from the previous cycle to the current detection cycle, by constructing a knowledge graph of the medical pure water machine parameters, the associated information of the modifications made to the medical pure water machine by the suspicious users is obtained. According to the historical data of the medical pure water machine parameters, the degree of influence of each parameter on the water quality is analyzed, and the importance information of the modifications made to the medical pure water machine by the suspicious users is obtained. The associated information of the modifications made to the medical pure water machine by the suspicious users is represented by the correlation closeness coefficient, and the importance information of the modifications made to the medical pure water machine by the suspicious users is represented by the risk importance regression coefficient.

[0101] It should be noted that after finding the suspicious user and abnormal parameters, it is necessary to further search for other potentially abnormal modification operations based on the abnormal parameters of the suspicious user, identify other potential abnormal modifications, and reduce misjudgments or omissions caused by single parameter anomalies. This includes:

[0102] A single parameter anomaly may be caused by a user mistake or temporary adjustment at a certain moment in the operation process, but it may also affect other related parameters. Relying solely on the anomaly of a single parameter as the identification standard is likely to miss some systemic anomalies caused by the combined effects of multiple parameters. Through multi-dimensional analysis, the correlation between multiple parameters can be systematically captured and potential abnormal behavior can be identified.

[0103] Abnormal modification of a single parameter does not necessarily lead directly to a malfunction or water quality problem, but is part of a systemic problem. Examining the interrelationships between multiple parameters can help us identify these potential systemic problems.

[0104] By comprehensively analyzing the operational behavior of suspicious users, potential equipment failures and anomalies can be predicted in advance, allowing maintenance or adjustments to be performed in advance. This proactive fault prevention strategy is more cost-effective than post-processing, reducing equipment downtime and improving equipment availability and stability.

[0105] The logic for obtaining the correlation closeness coefficient is as follows: constructing a knowledge graph of medical water purifier parameters, wherein the basic unit in the knowledge graph of medical water purifier parameters is a node, a node represents an entity, and represents the parameters of the medical water purifier. Nodes are connected by edges, which represent the relationship between the medical water purifier parameters. The edges between nodes have a weight value, which represents the strength of the relationship between the nodes.

[0106] Obtain the parameters of the medical pure water machine modified by the suspicious user each time, and represent the parameters of the medical pure water machine modified by the suspicious user each time as a set: XG g ={GB1, GB2, GB3,..., GB H}, where g = 1, 2, 3, ..., G, G is a positive integer, g is the number of times the suspected user has modified the medical pure water machine from the previous cycle to the current detection cycle, GB1, GB2, GB3, ..., GB H For each parameter modified by the suspicious user, the distance between the abnormal parameters of the medical pure water machine and the parameters modified by the suspicious user is determined through the shortest path analysis in the graph algorithm, and the distance between the abnormal parameters of the medical pure water machine and the parameters modified by the suspicious user is marked as: JL h ;

[0107] Calculate the correlation closeness coefficient, the calculation formula is: Among them, JM g The correlation closeness coefficient of the medical pure water machine modified by the suspicious user for the gth time, h = 1, 2, 3, ..., H, H is a positive integer, and h is the number of the medical pure water machine parameter modified for the gth time.

[0108] As can be seen from the formula, the smaller the correlation closeness coefficient, the more likely there is a problem with the operation, indicating that the suspicious user may not only modify abnormal parameters, but may also have illegally modified parameters in the operation.

[0109] The logic for obtaining the risk important regression coefficient is as follows: determining the parameters of the medical water purifier, collecting the historical data of the parameters of the medical water purifier, and obtaining the water quality data corresponding to the parameters of each medical water purifier, constructing a logistic risk important regression model through a logistic regression model, generating a risk important regression coefficient, and quantifying the importance of each parameter through the risk important regression coefficient;

[0110] It should be noted that the parameters of the medical pure water machine include flow rate, temperature, pressure, conductivity and pH value. By setting the parameters of the medical pure water machine, the water quality of the medical pure water machine can be affected.

[0111] The expression of the Logistic risk important regression model is: Wherein, HG is the risk importance regression coefficient of the Logistic risk importance regression model, CS1, CS2, CS3, …, CS B is the numerical value of all parameters of the medical pure water machine, λ1, λ2, λ3, …, λ B is the regression coefficient, indicating the influence degree of each parameter of the medical pure water machine on the water quality risk;

[0112] Obtain the parameters of the medical pure water machine after each operation of the suspicious user from the last cycle to the current detection cycle, and bring the parameters of the medical pure water machine after each operation of the suspicious user from the last cycle to the current detection cycle into the Logistic risk importance regression model, obtain the risk importance regression coefficient of the suspicious user after each operation of the medical pure water machine from the last cycle to the current detection cycle, and mark the risk importance regression coefficient of the suspicious user after each operation of the medical pure water machine from the last cycle to the current detection cycle as HG g .

[0113] As can be seen from the formula, the greater the risk importance regression coefficient of each operation of the suspicious user, the greater the influence of the operation on the water quality risk of the medical pure water machine.

[0114] Through comprehensive analysis of the association information and importance information modified by the suspicious user on the medical pure water machine, the association tightness coefficient and the risk importance regression coefficient are weighted and calculated to construct a modification evaluation model to generate a modification evaluation coefficient. The calculation formula of the modification evaluation coefficient is: Wherein, pg g is the modification evaluation coefficient of the gth modification of the suspicious user, γ1 and γ2 are proportional coefficients of the association tightness coefficient and the risk importance regression coefficient respectively, and γ1 and γ2 are both greater than 0.

[0115] As can be seen from the formula, the smaller the association tightness coefficient and the greater the risk importance regression coefficient, the greater the modification evaluation coefficient, indicating that the modification operation of the suspicious user on the medical pure water machine this time may have a problem, otherwise, the greater the association tightness coefficient and the smaller the risk importance regression coefficient, the smaller the modification evaluation coefficient, indicating that the modification operation of the suspicious user on the medical pure water machine this time may not have a problem.

[0116] Set the modification evaluation coefficient threshold, compare the modification evaluation coefficient of each modification of the suspicious user from the last cycle to the current detection cycle with the modification evaluation coefficient threshold, if the modification evaluation coefficient is greater than the modification evaluation coefficient threshold, mark the modification, indicating that this modification has a problem, which is used for professional staff to track potential parameter modification problems, if the modification evaluation coefficient is less than the modification evaluation coefficient threshold, do not mark the modification, indicating that this modification is normal.

[0117] The present invention periodically identifies abnormal parameters of medical water purifiers, conducts in-depth analysis based on the user's modification behavior, and then identifies potential abnormal modification records and ultimately determines the suspicious user, which helps to improve the operational safety of medical water purifiers, timely discover potential abnormal operations, and improve the efficiency of audit tracking.

[0118] Example 2

[0119] like Figure 2 The present invention provides a schematic structural diagram of a medical water purifier audit tracking system, which specifically includes an anomaly recognition module, a data acquisition module, a user evaluation module, and a record audit evaluation module, with signal connections between the modules.

[0120] The abnormality identification module is used to periodically monitor the operating status of the medical water purifier and identify abnormal parameters that occur in the current cycle;

[0121] The data collection module is used to collect and modify the user's historical operation information and historical habit information, and collect the associated information and importance information of the medical pure water machine modification;

[0122] The user evaluation module is used to build a user evaluation model, generate a user evaluation coefficient, and compare the user evaluation coefficient with the user evaluation coefficient threshold to identify suspicious users;

[0123] The record audit assessment module is used to build a modification assessment model, generate a modification assessment coefficient, and compare the modification assessment coefficient of the suspicious user with the modification assessment coefficient threshold to determine the potential abnormal modification records of the suspicious user.

[0124] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0125] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0126] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0127] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0128] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0130] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0131] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A medical water purifier audit tracking method, characterized in that: Specifically include the following steps: S1: Periodically identify abnormal parameters of the medical water purifier. When abnormal parameters of the medical water purifier are identified, retrieve the users who modified the abnormal parameters from the previous cycle to the current detection cycle; S2: By clustering the time points at which the user modified the water purifier, and comparing the similarity of the operating characteristics between the users, the user's historical operation information is determined. Based on the time series analysis of the abnormal parameters modified by the user, the user's historical habit information is determined. S3: Identify the suspected user through comprehensive analysis of the user's historical operation information and historical habit information, and extract the relevant information and importance information of each modification made by the suspected user to the medical pure water machine; S4: Determine the potential abnormal modification records of the suspicious user based on the associated information and importance information of the suspicious user's modification of the medical pure water machine; The historical operation information of the modified user is represented by the cluster center time distance difference coefficient and the operation feature similarity coefficient; The logic for obtaining the time-distance difference coefficient of the cluster center is as follows: obtaining the historical operation records of users who modified abnormal parameters from the previous cycle to the current detection cycle, marking the users who modified abnormal parameters from the previous cycle to the current detection cycle as modifying users, determining the time points when each modifying user modified the abnormal parameters of the medical pure water machine parameters, using the DBSCAN clustering algorithm to divide the operation time points of different modifying users, and setting the neighborhood radius to define the maximum distance between time points and the minimum number of points to determine the minimum number of time points required for a cluster. Areas with less than the minimum number of points are regarded as noise; Perform DBSCAN clustering on the modification time points of each user in a day to obtain clusters of abnormal parameters of the medical water purifier. Each cluster represents a regular time period of the modification user's operation, and the noise represents some abnormal operation time points. The cluster center of the cluster is approximated by calculating the mean of all time points in the cluster, and the cluster center of each cluster is marked as: SJ n , where n is the number of clusters after DBSCAN clustering; According to the operation time point of the modified user, determine whether there is a noise point in the modified user, and mark the noise point of the modified user as: Where i = 1, 2, 3, ..., I, where I is a positive integer and I is the number of the user who modified the abnormal parameters from the previous cycle to the current detection cycle; m = 1, 2, 3, ..., M, where M is a positive integer and m is the number of the noise point where the i-th modifying user exists; According to the noise point of the modified user, the minimum time difference between the noise point of the modified user and the cluster center of each cluster is determined, and the minimum time difference between the noise point of the modified user and the cluster center of each cluster is marked as: in, If there is no noise point for the modified user, the minimum time difference between the noise point of the modified user and the cluster center of each cluster is set to 0; Eliminate the noise points of user modification and mark the user modification time points according to the user modification time points: Where j = 1, 2, 3, ..., J, where J is a positive integer and j is the number of the i-th operation time point of the user modification. The minimum time difference between the operation time point of the user modification and the cluster center of each cluster is determined, and the minimum time difference between the operation time point of the user modification and the cluster center of each cluster is marked as: in, Calculate the cluster center time distance difference coefficient, the calculation formula is: Among them, cy i is the time distance difference coefficient of the cluster center of the i-th modified user.

2. A medical water purifier audit tracking method according to claim 1, characterized in that ,Periodically identify abnormal parameters of medical water purifiers, including: By using a card reader in combination with read-write cards of different authority levels to set equipment parameters, and by periodically conducting quality inspections on the medical pure water machine, the safety of the medical pure water machine is ensured. When problems arise in the quality inspection of the medical pure water machine, further inspections are conducted to determine the parameter settings of the medical pure water machine, and the abnormal parameter settings of the medical pure water machine are obtained. Based on the abnormal parameter settings of the medical pure water machine, users who have modified the abnormal parameters from the previous cycle to the current inspection cycle are retrieved.

3. A medical water purifier audit tracking method according to claim 2, characterized in that: The logic for obtaining the operation feature similarity coefficient is: Determine the operation characteristics of the modifying user. The operation characteristics of the modifying user include modification frequency, modification amplitude, and operation duration. The modification frequency is determined by determining the number of times the abnormal parameters are modified from the previous cycle to the current detection cycle. The modification amplitude is determined by recording the change amplitude of each abnormal parameter modification by the modifying user and determining the maximum amplitude of the abnormal parameter modification by the modifying user. The maximum amplitude of the abnormal parameter modification by the modifying user is used as the modification amplitude in the operation characteristics. The operation duration is determined by recording the average duration of each modification by the modifying user. The operation characteristics of the i-th modifying user are represented by a vector: X i =(pl i ,fd i ,sc i ); The similarity between the modifying user and the operation feature vectors of other modifying users is calculated by cosine similarity. The calculation formula is: Among them, XS ik is the similarity between the operation features of the i-th modifying user and the k-th other modifying user, X k is the operation feature vector of other modifying users except the i-th modifying user, k=1, 2, 3, ..., K, K is a positive integer, and k is the number of other modifying users except the i-th modifying user; Calculate the operation feature similarity coefficient, the calculation formula is: Among them, TZXS i is the operation feature similarity coefficient.

4. A medical water purifier audit tracking method according to claim 3, characterized in that: Determine and modify the user's historical habit information, including: Modify the user's historical habit information through the operation characteristic stability coefficient; The logic for obtaining the stability coefficient of the operation characteristic is as follows: according to the operation of modifying the abnormal parameters of the medical pure water machine by the user in the previous cycle, the difference of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle is obtained, and the difference of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle is marked as: CZ r , where r = 1, 2, 3, ..., R, R is a positive integer, and r is the number of each time the user modified the medical pure water machine in the previous cycle; According to the difference in the value of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle, the time series of the difference in the value of the modification of the abnormal parameters of the user in the previous cycle is determined, and the ARMA model is trained based on the time series data to determine the orders P and Q of the ARMA model. The difference in the value of each modification of the abnormal parameters of the medical pure water machine by the user in the previous cycle is used to predict the difference in the value of each modification of the abnormal parameters of the medical pure water machine by the user from the previous cycle to the current cycle; The ARMA model expression obtained by training is: Among them, α is a constant term, which represents the offset of the time series, p = 1, 2, 3, ..., P, p is the number of the order P of the ARMA model, q = 1, 2, 3, ..., Q, q is the number of the order Q of the ARMA model, φ p is the coefficient of the MA part in the ARMA model, which represents the relationship with the residuals of the past p moments, θ q is the coefficient of the AR part in the ARMA model, which represents the relationship with the value of the past q moments, ε r is the white noise term; The ARMA model obtained through training predicts the difference between the user's modification of the abnormal parameters of the medical pure water machine from the previous cycle to the current cycle. The expression is: Among them, d R+t To predict the difference in the user's t-th modification of the abnormal parameters of the medical pure water machine, d R+t-p is the known observation value of the time series data, ε R+t-q is the prediction error of time series data, t = 1, 2, 3, ..., T, T is a positive integer, and t is the number of modified differences of the required prediction anomaly parameters; Obtain the difference in the actual modification of the abnormal parameters of the medical pure water machine by the modifying user from the previous cycle to the current detection cycle, and mark the difference in the actual modification of the abnormal parameters of the medical pure water machine by the modifying user from the previous cycle to the current detection cycle as: DQ t ; By calculating the variance of the residual error of the modified user's abnormal parameter difference, the operating characteristic stability coefficient is determined, and the operating characteristic stability coefficient of the i-th modified user is marked as: WD i ,in, 5. A medical water purifier audit tracking method according to claim 4, characterized in that: Identify suspicious users, including: By modifying the user's historical operation information and historical habit information through comprehensive analysis, the cluster center time distance difference coefficient, operation feature similarity coefficient and operation feature stability coefficient are weighted and calculated to build a user evaluation model and generate a user evaluation coefficient. The calculation formula of the user evaluation coefficient is: Among them, pg i is the user evaluation coefficient of the i-th modifying user, β1, β2, and β3 are the proportional coefficients of the cluster center time distance difference coefficient, the operation feature similarity coefficient, and the operation feature stability coefficient, respectively. β1, β2, and β3 are all greater than 0; Set a user evaluation coefficient threshold, compare the user evaluation coefficients of all modified users with the user evaluation coefficient threshold, and mark the modified users whose user evaluation coefficients are greater than the user evaluation coefficient threshold as suspicious users. If the user evaluation coefficient of the modified user is less than the user evaluation coefficient threshold, no mark will be made.

6. A medical water purifier audit tracking method according to claim 5, characterized in that: The relevant information of each modification made by the suspicious user to the medical pure water machine, including: The associated information of the suspected user's modification of the medical pure water machine is represented by the correlation closeness coefficient; The logic for obtaining the correlation closeness coefficient is as follows: constructing a knowledge graph of medical water purifier parameters, wherein the basic unit in the knowledge graph of medical water purifier parameters is a node, a node represents an entity, and represents the parameters of the medical water purifier. Nodes are connected by edges, which represent the relationship between the medical water purifier parameters. The edges between nodes have a weight value, which represents the strength of the relationship between the nodes. Obtain the parameters of the medical pure water machine modified by the suspicious user each time, and represent the parameters of the medical pure water machine modified by the suspicious user each time as a set: XG g ={GB1, GB2, GB3,..., GB H }, where g = 1, 2, 3, ..., G, G is a positive integer, g is the number of times the suspected user has modified the medical pure water machine from the previous cycle to the current detection cycle, GB1, GB2, GB3, ..., GB H For each parameter modified by the suspicious user, the distance between the abnormal parameters of the medical pure water machine and the parameters modified by the suspicious user is determined through the shortest path analysis in the graph algorithm, and the distance between the abnormal parameters of the medical pure water machine and the parameters modified by the suspicious user is marked as: JL h ; Calculate the correlation closeness coefficient, the calculation formula is: Among them, JM g The correlation closeness coefficient of the medical pure water machine modified by the suspicious user for the gth time, h = 1, 2, 3, ..., H, H is a positive integer, and h is the number of the medical pure water machine parameter modified for the gth time.

7. A medical water purifier audit tracking method according to claim 6, characterized in that: The importance of each modification made by the user to the medical water purifier includes: The importance of the modification of the medical water purifier by the doubtful user is expressed by the risk important regression coefficient; The logic for obtaining the risk important regression coefficient is as follows: determining the parameters of the medical water purifier, collecting the historical data of the parameters of the medical water purifier, and obtaining the water quality data corresponding to the parameters of each medical water purifier, constructing a logistic risk important regression model through a logistic regression model, generating a risk important regression coefficient, and quantifying the importance of each parameter through the risk important regression coefficient; The expression of the Logistic risk important regression model is: Among them, HG is the risk important regression coefficient of the Logistic risk important regression model, CS1, CS2, CS3, ..., CS B are the values ​​of all parameters of the medical water purifier, λ1, λ2, λ3, ..., λ B is the regression coefficient, which indicates the influence of each medical water purifier parameter on water quality risk; Obtain the parameters of the medical pure water machine after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle, and bring the parameters of the medical pure water machine after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle into the Logistic risk important regression model to obtain the risk important regression coefficient after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle, and mark the risk important regression coefficient after each operation of the medical pure water machine by the suspicious user from the previous cycle to the current detection cycle as: HG g .

8. A medical water purifier audit tracking method according to claim 7, characterized in that: Identify potential abnormal modification records of suspicious users, including: Through a comprehensive analysis of the relevance and importance information of the suspicious users' modifications to the medical pure water machine, the correlation closeness coefficient and the risk important regression coefficient are weighted and calculated to construct a modification evaluation model and generate a modification evaluation coefficient. The calculation formula of the modification evaluation coefficient is: Among them, pg g is the modification evaluation coefficient of the g-th modification by the suspicious user, γ1 and γ2 are the proportional coefficients of the correlation closeness coefficient and the risk important regression coefficient, respectively, and γ1 and γ2 are both greater than 0; Set a modification evaluation coefficient threshold and compare the modification evaluation coefficient of each modification made by the suspicious user from the previous cycle to the current detection cycle with the modification evaluation coefficient threshold. If the modification evaluation coefficient is greater than the modification evaluation coefficient threshold, the modification is marked; if the modification evaluation coefficient is less than the modification evaluation coefficient threshold, the modification is not marked.

9. A medical water purifier audit tracking system, used to implement the medical water purifier audit tracking method according to any one of claims 1 to 8, characterized in that: It includes anomaly identification module, data acquisition module, user evaluation module and record audit evaluation module, and signal connections between modules; The abnormality identification module is used to periodically monitor the operating status of the medical water purifier and identify abnormal parameters that occur in the current cycle; The data collection module is used to collect and modify the user's historical operation information and historical habit information, and collect the associated information and importance information of the medical pure water machine modification; The user evaluation module is used to build a user evaluation model, generate a user evaluation coefficient, and compare the user evaluation coefficient with the user evaluation coefficient threshold to identify suspicious users; The record audit assessment module is used to build a modification assessment model, generate a modification assessment coefficient, and compare the modification assessment coefficient of the suspicious user with the modification assessment coefficient threshold to determine the potential abnormal modification records of the suspicious user.

Images