POS machine service inspection system

Abstract

The invention discloses a POS machine service inspection system, which comprises an anti-fraud security certification module, a software and hardware inspection module, a signal inspection module and a risk division decision module, and is characterized in that the software and hardware inspection module is used for collecting hardware and software monitoring data of a plurality of POS machines to be inspected, constructing a first inspection path of the plurality of POS machines to be inspected, and sending the first inspection path to the signal inspection module; obtaining first inspection data of a plurality of POS machines to be inspected, and generating risk feature information; the signal inspection module is used for carrying out signal path analysis on the POS machine to be inspected, determining whether the POS machine to be inspected has signal transmission abnormity and a component corresponding to the transmission abnormity, and generating risk feature information if the POS machine to be inspected has the signal transmission abnormity and the component corresponding to the transmission abnormity; according to the method and the system, a rapid priority inspection path and a differentiated signal inspection scheme are constructed during POS machine software and hardware inspection, so that risk investigation of field staff inspection can be rapidly realized, differentiated inspection can be performed aiming at various different inspection problems of the POS machine, and the inspection efficiency is improved.

Description

Technical Field

[0001] The invention belongs to the field of field staff inspection of POS machine business, and specifically refers to a POS machine business inspection system. Background Art

[0002] With the rapid development of mobile payment and e-commerce, POS machines, as key tools for transaction settlement, are crucial to the operation of merchants due to their stability and security. However, the post-maintenance and management of POS machines face many challenges, such as the widespread distribution of equipment, heavy inspection tasks, and untimely information feedback.

[0003] The main problems with the field staff inspection system for POS business are: when problems occur at several POS machines, a fast inspection path cannot be built, inspection tasks are not handled according to the priority of the problem, and the inspection efficiency is low. In addition, POS machine signal inspection is time-consuming and complicated, and efficient inspection problem troubleshooting cannot be quickly achieved. Summary of the invention

[0004] In order to solve the problems in the above-mentioned prior art that when problems occur in several POS machines, a fast inspection path cannot be constructed, inspection tasks are not processed according to the priority of the problems, the inspection efficiency is low, and in addition, the POS machine signal inspection is time-consuming and complicated, and efficient inspection problem troubleshooting cannot be quickly achieved, the present invention proposes a POS machine business inspection system to improve the above-mentioned problems.

[0005] The specific application is as follows:

[0006] A POS machine business inspection system, including an anti-fraud security authentication module, a software and hardware inspection module, a signal inspection module and a risk-based decision-making module;

[0007] The anti-fraud security authentication module is used for the system to support customized inspection forms and QR codes. Field staff can take photos on site and attach watermarks to record the date, location and personnel of the inspection to prevent fraud by field staff. Inspection initialization security authentication is performed by scanning the QR code specified by the system.

[0008] The hardware and software inspection module is used to collect hardware and software monitoring data of several POS machines to be inspected, confirm whether the POS machines to be inspected are in normal operating mode through a first analysis result, obtain location information of several POS machines to be inspected in abnormal operating mode, construct a first inspection path for several POS machines to be inspected, and then obtain first inspection data of several POS machines to be inspected, and generate risk feature information;

[0009] The signal inspection module is used to perform signal path analysis on the POS machine to be inspected, determine whether the POS machine to be inspected has signal transmission anomalies and components corresponding to the transmission anomalies, and if so, generate risk feature information;

[0010] The risk-based decision-making module is used to add a historical risk processing data set with risk processing measure labels to train the preset risk processing model, obtain a pre-trained risk processing model after the risk processing model training, input the risk feature information into the pre-trained risk processing model, obtain a corresponding risk decision report, and feed back the risk decision report to the field inspection personnel.

[0011] Furthermore, the anti-fraud security authentication module performs inspection and initialization security authentication by scanning the QR code specified by the system. The specific implementation process includes:

[0012] L1: The field inspection personnel obtain the encryption parameters sent by the server; wherein the encryption parameters sent by the server each time are different, and the encryption parameters correspond to the ID numbers corresponding to the field inspection personnel one by one;

[0013] L2: The field inspection personnel display a QR code that matches the encryption parameters, wherein the QR code is used to instruct the inspection device to decrypt the scanned QR code based on the encryption parameters sent by the server after being scanned. If the decryption is successful, the inspection initialization of the inspected POS machine is completed. If not, the inspection initialization of the inspected POS machine cannot be performed.

[0014] Further, the software and hardware inspection module includes an operation data acquisition submodule, a data analysis submodule, a software and hardware information acquisition submodule, a path construction submodule and a data acquisition submodule;

[0015] The operation data collection submodule is used to deploy monitoring equipment on the POS machine to be inspected, collect the operation parameters of the POS machine in real time, and obtain the monitoring data of each node of the POS machine. The monitoring data of each node includes hardware monitoring data, software system monitoring data and data security monitoring data;

[0016] The data analysis submodule is used to input the POS machine monitoring data into a state analysis model to obtain a first analysis result, wherein the first analysis result includes a normal mode and an abnormal mode;

[0017] The software and hardware information acquisition submodule is used to retrieve the historical inspection information and location information of the POS machines in several abnormal modes, wherein the historical inspection information includes a historical inspection time set;

[0018] A path construction submodule, used to construct an inspection path according to the historical inspection time set and the location information to generate a first inspection path;

[0019] A data acquisition submodule, used for performing inspections on the POS machines in the plurality of abnormal modes to acquire first inspection data;

[0020] The hardware monitoring data includes whether the appearance is intact, whether the buttons are sensitive, whether the printing device can work normally, and whether the connecting cables are loose or damaged;

[0021] The software system monitoring data includes checking whether the operating system is the latest version, whether there are any uninstalled security patches or system upgrade prompts, and running the built-in diagnostic program to test whether the software functions are normal. The running of the built-in diagnostic program to test whether the software functions are normal includes whether the card swiping is normal, whether the code scanning is normal, and whether the data upload is normal;

[0022] The data security monitoring data includes verifying whether the data backup mechanism of the POS system is effective, checking the firewall settings, login password strength, and whether the virus protection software status is abnormal.

[0023] Furthermore, the data analysis submodule is used to input the POS machine monitoring data into the state analysis model to obtain the first analysis result. The specific implementation process includes:

[0024] M1: In the first node of the POS machine to be inspected, collect historical POS machine monitoring data within a preset historical period;

[0025] M2: Acquire a POS machine monitoring data set, evaluate the historical POS machine monitoring data according to the POS machine monitoring data set, mark the evaluation results that meet the threshold as normal mode, and mark the evaluation results that do not meet the threshold as abnormal mode, and obtain the historical POS machine monitoring data analysis results;

[0026] M3: annotating the historical POS machine monitoring data analysis results to obtain a first constructed data set;

[0027] M4: construct a first POS machine state analysis model, and use the first constructed data set to perform deep learning on the first POS machine state analysis model to obtain the first POS machine state analysis model whose accuracy meets a threshold;

[0028] M5: Continue to construct corresponding status analysis models at other POS machine nodes respectively, and input the monitoring data of each POS machine node into the corresponding device status analysis model to obtain the first analysis results of N nodes.

[0029] Furthermore, the specific implementation process of the path construction submodule includes:

[0030] N1: According to the location information of several POS machines, obtain the coordinates of several POS machines, where the several coordinates are set as Q;

[0031] N2: constructing a path for the Q coordinates according to path constraints to generate a path construction result, wherein the path constraints are determined according to the remoteness of the location of a single POS terminal and its usage activity;

[0032] N3: Obtain the last inspection time and the number of historical inspections of the Q POS terminals according to the historical inspection time set;

[0033] N4: Prioritize the Q POS terminals according to the last inspection time and the number of historical inspections to obtain the priority of the POS terminals;

[0034] N5: Adjust the path construction result according to the priority level of the POS machine to obtain a first inspection path.

[0035] Further, the N2 specifically includes:

[0036] N21: randomly generating an initial first inspection path set according to the path constraint condition and the Q POS machine coordinates, wherein each initial first inspection path represents a type of solution, which is composed of the Q POS machine coordinates;

[0037] N22: Obtain the distance information between POS terminals, and construct a matching function based on the distance information, as shown in the following formula:

[0038]

[0039] Where P represents the matching degree, c represents the distance information between POS machines, n represents the number of pairs of matches among Q POS machines, and i represents any pair of POS machines;

[0040] N23: Calculate the initial first inspection path set according to the matching function, and generate an optimal path construction result according to the path calculation result.

[0041] Furthermore, the N23 specifically includes:

[0042] N231: randomly selecting a first inspection path calculation result from the path calculation results as the current optimal calculation result;

[0043] N232: Continue to randomly select a second inspection path calculation result from the path calculation results;

[0044] N233: Determine whether the calculation result of the second patrol path is greater than the calculation result of the first patrol path. If so, use the calculation result of the second patrol path as the optimal calculation result;

[0045] N234: Continue iterating and optimizing until a preset number of iterations is reached, and output the initial first inspection path corresponding to the final optimal calculation result as the optimal path construction result.

[0046] Furthermore, the signal inspection module specifically implements the following process:

[0047] K1: Decompose the signal path of the POS machine to be inspected, determine the signal path set, perform corresponding function analysis on the signal path set, and establish a signal-component relationship list;

[0048] K2: construct a signal processing unit, which is set in the POS machine and includes a plurality of signal processing sub-units, which correspond to the signal paths;

[0049] K3: Establishing a transmission relationship between the signal path set and the signal processing unit, connecting each signal path through each signal processing sub-unit, and receiving the signal sent by the POS machine;

[0050] K4: Based on the signal-component relationship list, set abnormal signal characteristics by collecting historical signal data;

[0051] K5: Perform feature comparison on the received signals in each signal processing subunit based on the abnormal signal feature, and locate the abnormal components in the POS terminal according to the signal comparison results.

[0052] Furthermore, the signal path of the POS terminal to be inspected is decomposed in K1 to determine the signal path set, including:

[0053] K11: Perform signal path analysis on the POS machine to be inspected and determine the signal path devices, wherein the signal path devices include wireless network devices and IoT sensor devices;

[0054] K12: Based on the signal path information, perform signal path feature analysis to determine the range and connection mode of each signal path;

[0055] K13: Analyze the signal path characteristics of the POS machine to be inspected to determine the signal path type and signal path range;

[0056] K14: performing credibility matching based on each signal path range and connection mode, signal path type and signal path range, to obtain the signal path set, where the signal path set is a signal path that satisfies a matching relationship;

[0057] The K4 specifically includes:

[0058] K41: Determine the signal type of each component based on the signal-component relationship list;

[0059] K42: Based on the signal type of each component, collect historical signal data of the component and determine an abnormal signal set;

[0060] K43: Use deep learning model to identify abnormal signal sets and identify abnormal signal features;

[0061] The specific implementation process of K43 is as follows:

[0062] K431: Collect 100,000 pieces of historical signal data of components, where the historical signal data of components are divided into an abnormal signal set and a non-abnormal signal set;

[0063] K432: Divide 100,000 records into training set, test set and validation set in a ratio of 8:1:1; the data in the training set contains all abnormal signal set information, the data in the test set contains both abnormal signal set and non-abnormal signal set information, and the data in the validation set contains both abnormal signal set information and non-abnormal signal set information;

[0064] K433: Use annotation tools to annotate the data in the training set and generate an XML file;

[0065] K434: After extracting features using a convolutional neural network algorithm, the deep learning model trains the data in the training set one by one;

[0066] K435: The deep learning model recognizes the data in the test set one by one and obtains the accuracy and error rate;

[0067] K436: If the accuracy and error rate do not meet the requirements, proceed to step K437; if the accuracy and error rate meet the requirements, the deep learning model recognizes the data in the validation set one by one, and if the accuracy and error rate meet the requirements, the training ends; if the accuracy and error rate do not meet the requirements, proceed to step K437;

[0068] K437: After manually adjusting the recognition weight coefficient of the deep learning model, the data in the validation set are recognized one by one. If the recognition accuracy and error rate meet the requirements, the training ends. If not, the recognition weight coefficient is adjusted again and the data in the validation set are recognized one by one again. After each recognition of the data in the validation set, the recognition record is deleted until the recognition accuracy and error rate meet the requirements and the training ends.

[0069] K438: The trained deep learning model is used to identify abnormal signal features. If the input signal is found to be an abnormal feature, the signal is set as an abnormal feature.

[0070] The K5 specifically includes:

[0071] K51: extract the abnormal signal feature of the signal processing subunit, and if the abnormal signal feature is a strength feature, obtain the abnormal strength interval;

[0072] K52: Acquire a receiving signal path of the signal processing subunit, and construct a signal strength waveform based on the receiving signal path;

[0073] K53: Adapt the signal strength waveform to the abnormal coordinates constructed by the abnormal strength interval, obtain the abnormal trigger point, locate the position of the abnormal components in the POS machine, and generate risk feature information.

[0074] Furthermore, the specific implementation process of the risk sharing decision module is as follows:

[0075] T1: Train the preset risk treatment model using the historical risk treatment data set with risk treatment measure labels added to obtain a corresponding set of risk treatment model training losses;

[0076] T2: setting the optimal risk processing model training loss as the loss parameter of the preset risk processing model to obtain a pre-trained risk processing model after the risk processing model training;

[0077] T3: inputting the risk characteristic information into the pre-trained risk processing model to obtain the risk processing measures output by the pre-trained risk processing model;

[0078] T4: According to the preset risk treatment measure scoring rules and the risk treatment measures, determine the score of the risk treatment measures and obtain a corresponding risk decision report.

[0079] The beneficial effects of a POS machine business inspection system of the present invention are as follows:

[0080] The present invention uses an anti-fraud security authentication module to support custom inspection forms and QR codes in the system. Field staff take photos on site and attach watermarks to record the date, location and personnel of the inspection to prevent field staff from committing fraud. Inspection initialization security authentication is performed by scanning the QR code specified by the system; the hardware and software inspection module is used to collect hardware and software monitoring data of several POS machines to be inspected, and confirm whether the POS machines to be inspected are in normal operating mode through a first analysis result, obtain the location information of several POS machines to be inspected in abnormal normal operating mode, construct the first inspection path of several POS machines to be inspected, and then obtain the first inspection data of several POS machines to be inspected, and generate risk feature information; the signal inspection module is used to perform signal path analysis on the POS machines to be inspected, determine The POS machine to be inspected is checked to see whether there are signal transmission anomalies and components corresponding to the transmission anomalies. If so, risk feature information is generated; a risk decision-making module is used to add a historical risk processing data set with risk processing measure labels to train the preset risk processing model, obtain a pre-trained risk processing model after the risk processing model training, input the risk feature information into the pre-trained risk processing model, obtain a corresponding risk decision report, and feed back the risk decision report to the field inspection personnel; the present invention can quickly realize the risk investigation of field personnel inspections by constructing a fast priority inspection path and a solution for distinguishing signal inspections when inspecting the software and hardware of the POS machine, and can distinguish and inspect various different inspection problems of the POS machine to improve the inspection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 This is a system architecture diagram of a POS machine business inspection system of the present invention;

[0082] Figure 2 This is a diagram of the software and hardware inspection modules of a POS machine business inspection system of the present invention.

[0083] Among them, 1-anti-fraud security authentication module, 2-software and hardware inspection module, 21-operation data collection sub-module, 22-data analysis sub-module, 23-software and hardware information acquisition sub-module, 24-path construction sub-module, 25-data acquisition sub-module, 3-signal inspection module, 4-risk decision-making module. DETAILED DESCRIPTION

[0084] The technical solution of the present invention will be described clearly and completely below 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 creative work are within the scope of protection of the present invention.

[0085] Example 1

[0086] The system includes anti-fraud security authentication module 1, software and hardware inspection module 2, signal inspection module 3 and risk division decision module 4. Figure 1 and Figure 2 ;

[0087] The anti-fraud security authentication module 1 is used for the system to support customized inspection forms and QR codes. Field staff take photos on site and attach watermarks to record the date, location and personnel of the inspection to prevent field staff from cheating. Inspection initialization security authentication is performed by scanning the QR code specified by the system;

[0088] The hardware and software inspection module 2 is used to collect hardware and software monitoring data of several POS machines to be inspected, confirm whether the POS machines to be inspected are in normal operating mode through a first analysis result, obtain location information of several POS machines to be inspected in abnormal operating mode, construct a first inspection path for several POS machines to be inspected, and then obtain first inspection data of several POS machines to be inspected, and generate risk feature information;

[0089] The signal inspection module 3 is used to perform signal path analysis on the POS machine to be inspected, determine whether the POS machine to be inspected has signal transmission anomalies and components corresponding to the transmission anomalies, and if so, generate risk feature information;

[0090] The risk-based decision module 4 is used to add a historical risk processing data set with risk processing measure labels to train the preset risk processing model, obtain a pre-trained risk processing model after the risk processing model training, input the risk feature information into the pre-trained risk processing model, obtain a corresponding risk decision report, and feed back the risk decision report to the field inspection personnel.

[0091] Furthermore, the anti-fraud security authentication module 1 performs inspection initialization security authentication by scanning the QR code specified by the system. The specific implementation process includes:

[0092] L1: The field inspection personnel obtain the encryption parameters sent by the server; wherein the encryption parameters sent by the server each time are different, and the encryption parameters correspond to the ID numbers corresponding to the field inspection personnel one by one;

[0093] L2: The field inspection personnel display a QR code that matches the encryption parameters, wherein the QR code is used to instruct the inspection device to decrypt the scanned QR code based on the encryption parameters sent by the server after being scanned. If the decryption is successful, the inspection initialization of the inspected POS machine is completed. If not, the inspection initialization of the inspected POS machine cannot be performed.

[0094] Specifically, the anti-fraud security authentication module is not limited to the use of QR codes for inspection initialization security authentication, but can also include the use of Ukey card insertion for inspection initialization authentication. At the same time, the authentication requires uploading the personal photo on the server for synchronous authentication.

[0095] Further, the software and hardware inspection module 2 includes an operation data collection submodule 21, a data analysis submodule 22, a software and hardware information acquisition submodule 23, a path construction submodule 24 and a data acquisition submodule 25;

[0096] The operation data collection submodule 21 is used to deploy monitoring equipment at the POS machine to be inspected, collect the operation parameters of the POS machine in real time, and obtain the monitoring data of each node of the POS machine. The monitoring data of each node includes hardware monitoring data, software system monitoring data and data security monitoring data;

[0097] The data analysis submodule 22 is used to input the POS machine monitoring data into a state analysis model to obtain a first analysis result, wherein the first analysis result includes a normal mode and an abnormal mode;

[0098] The software and hardware information acquisition submodule 23 is used to retrieve the historical inspection information and location information of the POS machines in several abnormal modes, wherein the historical inspection information includes a set of historical inspection time;

[0099] A path construction submodule 24, configured to construct an inspection path according to the historical inspection time set and the location information, and generate a first inspection path;

[0100] The data acquisition submodule 25 is used to perform inspections on the POS machines in the plurality of abnormal modes and acquire first inspection data;

[0101] The hardware monitoring data includes whether the appearance is intact, whether the buttons are sensitive, whether the printing device can work normally, and whether the connecting cables are loose or damaged;

[0102] The software system monitoring data includes checking whether the operating system is the latest version, whether there are any uninstalled security patches or system upgrade prompts, and running the built-in diagnostic program to test whether the software functions are normal. The running of the built-in diagnostic program to test whether the software functions are normal includes whether the card swiping is normal, whether the code scanning is normal, and whether the data upload is normal;

[0103] The data security monitoring data includes verifying whether the data backup mechanism of the POS system is effective, checking the firewall settings, login password strength, and whether the virus protection software status is abnormal.

[0104] Furthermore, the data analysis submodule 22 is used to input the POS machine monitoring data into the state analysis model to obtain the first analysis result. The specific implementation process includes:

[0105] M1: In the first node of the POS machine to be inspected, collect historical POS machine monitoring data within a preset historical period;

[0106] M2: Acquire a POS machine monitoring data set, evaluate the historical POS machine monitoring data according to the POS machine monitoring data set, mark the evaluation results that meet the threshold as normal mode, and mark the evaluation results that do not meet the threshold as abnormal mode, and obtain the historical POS machine monitoring data analysis results;

[0107] M3: annotating the historical POS machine monitoring data analysis results to obtain a first constructed data set;

[0108] M4: construct a first POS machine state analysis model, and use the first constructed data set to perform deep learning on the first POS machine state analysis model to obtain the first POS machine state analysis model whose accuracy meets a threshold;

[0109] M5: Continue to construct corresponding status analysis models at other POS machine nodes respectively, and input the monitoring data of each POS machine node into the corresponding device status analysis model to obtain the first analysis results of N nodes.

[0110] Furthermore, the specific implementation process of the path construction submodule 24 includes:

[0111] N1: According to the location information of several POS machines, obtain the coordinates of several POS machines, where the several coordinates are set as Q;

[0112] N2: constructing a path for the Q coordinates according to path constraints to generate a path construction result, wherein the path constraints are determined according to the remoteness of the location of a single POS terminal and its usage activity;

[0113] N3: Obtain the last inspection time and the number of historical inspections of the Q POS terminals according to the historical inspection time set;

[0114] N4: Prioritize the Q POS terminals according to the last inspection time and the number of historical inspections to obtain the priority of the POS terminals;

[0115] N5: Adjust the path construction result according to the priority level of the POS machine to obtain a first inspection path.

[0116] Further, the N2 specifically includes:

[0117] N21: randomly generating an initial first inspection path set according to the path constraint condition and the Q POS machine coordinates, wherein each initial first inspection path represents a type of solution, which is composed of the Q POS machine coordinates;

[0118] N22: Obtain the distance information between POS terminals, and construct a matching function based on the distance information, as shown in the following formula:

[0119]

[0120] Where P represents the matching degree, c represents the distance information between POS machines, n represents the number of pairs of matches among Q POS machines, and i represents any pair of POS machines;

[0121] N23: Calculate the initial first inspection path set according to the matching function, and generate an optimal path construction result according to the path calculation result.

[0122] Furthermore, the N23 specifically includes:

[0123] N231: randomly selecting a first inspection path calculation result from the path calculation results as the current optimal calculation result;

[0124] N232: Continue to randomly select a second inspection path calculation result from the path calculation results;

[0125] N233: Determine whether the calculation result of the second patrol path is greater than the calculation result of the first patrol path. If so, use the calculation result of the second patrol path as the optimal calculation result;

[0126] N234: Continue iterating and optimizing until a preset number of iterations is reached, and output the initial first inspection path corresponding to the final optimal calculation result as the optimal path construction result.

[0127] Furthermore, the signal inspection module 3 specifically implements the following process:

[0128] K1: Decompose the signal path of the POS machine to be inspected, determine the signal path set, perform corresponding function analysis on the signal path set, and establish a signal-component relationship list;

[0129] K2: construct a signal processing unit, which is set in the POS machine and includes a plurality of signal processing sub-units, which correspond to the signal paths;

[0130] K3: Establishing a transmission relationship between the signal path set and the signal processing unit, connecting each signal path through each signal processing sub-unit, and receiving the signal sent by the POS machine;

[0131] K4: Based on the signal-component relationship list, set abnormal signal characteristics by collecting historical signal data;

[0132] K5: Perform feature comparison on the received signals in each signal processing subunit based on the abnormal signal feature, and locate the abnormal components in the POS terminal according to the signal comparison results.

[0133] Specifically, during the signal inspection process, field staff do not need to manually inspect equipment. They only need to perform initialization authentication on the anti-fraud security authentication module and then select the POS device to perform the signal inspection function to fully automate the signal inspection. The inspection results are accurate and efficient, which solves the problem that the POS signal inspection in the prior art is time-consuming and complicated, and cannot quickly and efficiently implement inspection problem troubleshooting.

[0134] Furthermore, the signal path of the POS terminal to be inspected is decomposed in K1 to determine the signal path set, including:

[0135] K11: Perform signal path analysis on the POS machine to be inspected and determine the signal path devices, wherein the signal path devices include wireless network devices and IoT sensor devices;

[0136] K12: Based on the signal path information, perform signal path feature analysis to determine the range and connection mode of each signal path;

[0137] K13: Analyze the signal path characteristics of the POS machine to be inspected to determine the signal path type and signal path range;

[0138] K14: performing credibility matching based on each signal path range and connection mode, signal path type and signal path range, to obtain the signal path set, where the signal path set is a signal path that satisfies a matching relationship;

[0139] The K4 specifically includes:

[0140] K41: Determine the signal type of each component based on the signal-component relationship list;

[0141] K42: Based on the signal type of each component, collect historical signal data of the component and determine an abnormal signal set;

[0142] K43: Use deep learning model to identify abnormal signal sets and identify abnormal signal features;

[0143] The specific implementation process of K43 is as follows:

[0144] K431: Collect 100,000 pieces of historical signal data of components, where the historical signal data of components are divided into an abnormal signal set and a non-abnormal signal set;

[0145] K432: Divide 100,000 records into training set, test set and validation set in a ratio of 8:1:1; the data in the training set contains all abnormal signal set information, the data in the test set contains both abnormal signal set and non-abnormal signal set information, and the data in the validation set contains both abnormal signal set information and non-abnormal signal set information;

[0146] K433: Use annotation tools to annotate the data in the training set and generate an XML file;

[0147] K434: After extracting features using a convolutional neural network algorithm, the deep learning model trains the data in the training set one by one;

[0148] K435: The deep learning model recognizes the data in the test set one by one and obtains the accuracy and error rate;

[0149] K436: If the accuracy and error rate do not meet the requirements, proceed to step K437; if the accuracy and error rate meet the requirements, the deep learning model recognizes the data in the validation set one by one, and if the accuracy and error rate meet the requirements, the training ends; if the accuracy and error rate do not meet the requirements, proceed to step K437;

[0150] K437: After manually adjusting the recognition weight coefficient of the deep learning model, the data in the validation set are recognized one by one. If the recognition accuracy and error rate meet the requirements, the training ends. If not, the recognition weight coefficient is adjusted again and the data in the validation set are recognized one by one again. After each recognition of the data in the validation set, the recognition record is deleted until the recognition accuracy and error rate meet the requirements and the training ends.

[0151] K438: The trained deep learning model is used to identify abnormal signal features. If the input signal is found to be an abnormal feature, the signal is set as an abnormal feature.

[0152] The K5 specifically includes:

[0153] K51: extract the abnormal signal feature of the signal processing subunit, and if the abnormal signal feature is a strength feature, obtain the abnormal strength interval;

[0154] K52: Acquire a receiving signal path of the signal processing subunit, and construct a signal strength waveform based on the receiving signal path;

[0155] K53: Adapt the signal strength waveform to the abnormal coordinates constructed by the abnormal strength interval, obtain the abnormal trigger point, locate the position of the abnormal components in the POS machine, and generate risk feature information.

[0156] Furthermore, the specific implementation process of the risk sharing decision module 4 is as follows:

[0157] T1: Train the preset risk treatment model using the historical risk treatment data set with risk treatment measure labels added to obtain a corresponding set of risk treatment model training losses;

[0158] T2: setting the optimal risk processing model training loss as the loss parameter of the preset risk processing model to obtain a pre-trained risk processing model after the risk processing model training;

[0159] T3: inputting the risk characteristic information into the pre-trained risk processing model to obtain the risk processing measures output by the pre-trained risk processing model;

[0160] T4: According to the preset risk treatment measure scoring rules and the risk treatment measures, determine the score of the risk treatment measures and obtain a corresponding risk decision report.

[0161] The beneficial effects of a POS machine business inspection system of the present invention are as follows:

[0162] The present invention uses an anti-fraud security authentication module to support custom inspection forms and QR codes in the system. Field staff take photos on site and attach watermarks to record the date, location and personnel of the inspection to prevent field staff from committing fraud. Inspection initialization security authentication is performed by scanning the QR code specified by the system; the hardware and software inspection module is used to collect hardware and software monitoring data of several POS machines to be inspected, and confirm whether the POS machines to be inspected are in normal operating mode through a first analysis result, obtain the location information of several POS machines to be inspected in abnormal normal operating mode, construct the first inspection path of several POS machines to be inspected, and then obtain the first inspection data of several POS machines to be inspected, and generate risk feature information; the signal inspection module is used to perform signal path analysis on the POS machines to be inspected, determine The POS machine to be inspected is checked to see whether there are signal transmission anomalies and components corresponding to the transmission anomalies. If so, risk feature information is generated; a risk decision-making module is used to add a historical risk processing data set with risk processing measure labels to train the preset risk processing model, obtain a pre-trained risk processing model after the risk processing model training, input the risk feature information into the pre-trained risk processing model, obtain a corresponding risk decision report, and feed back the risk decision report to the field inspection personnel; the present invention can quickly realize the risk investigation of field personnel inspections by constructing a fast priority inspection path and a solution for distinguishing signal inspections when inspecting the software and hardware of the POS machine, and can distinguish and inspect various different inspection problems of the POS machine to improve the inspection efficiency.

[0163] The present invention and its embodiments are described above, which is not restrictive. What is shown in the accompanying drawings is only one of the embodiments of the present invention, and the actual content is not limited to this. In short, if ordinary technicians in this field are inspired by it and do not deviate from the purpose of the invention, they can creatively design structural methods and embodiments similar to the technical scheme, which should all fall within the scope of protection of the present invention.

Claims

1. A POS machine business inspection system, characterized in that: It includes an anti-fraud security authentication module (1), a software and hardware inspection module (2), a signal inspection module (3) and a risk division decision module (4); The anti-fraud security authentication module (1) is used for the system to support customized inspection forms and QR codes. Field staff can take photos on site and attach watermarks to record the date, location and personnel of the inspection to prevent field staff from cheating. Inspection initialization security authentication is performed by scanning the QR code specified by the system. The hardware and software inspection module (2) is used to collect hardware and software monitoring data of a plurality of POS machines to be inspected, confirm whether the POS machines to be inspected are in normal operating mode through a first analysis result, obtain location information of a plurality of POS machines to be inspected in abnormal operating mode, construct a first inspection path for the plurality of POS machines to be inspected, and then obtain first inspection data of the plurality of POS machines to be inspected, and generate risk characteristic information; The signal inspection module (3) is used to perform signal path analysis on the POS machine to be inspected, determine whether the POS machine to be inspected has signal transmission anomalies and components corresponding to the transmission anomalies, and if so, generate risk feature information; The risk-based decision module (4) is used to add a historical risk processing data set with risk processing measure labels to perform risk processing model training on a preset risk processing model, obtain a pre-trained risk processing model after the risk processing model training, input the risk feature information into the pre-trained risk processing model, obtain a corresponding risk decision report, and feed back the risk decision report to the field inspection personnel.

2. A POS machine business inspection system according to claim 1, characterized in that: The anti-fraud security authentication module (1) performs inspection and initialization security authentication by scanning a two-dimensional code specified by the system. The specific implementation process includes: L1: The field inspection personnel obtain the encryption parameters sent by the server; wherein the encryption parameters sent by the server each time are different, and the encryption parameters correspond to the ID numbers corresponding to the field inspection personnel one by one; L2: The field inspection personnel display a QR code that matches the encryption parameters, wherein the QR code is used to instruct the inspection device to decrypt the scanned QR code based on the encryption parameters sent by the server after being scanned. If the decryption is successful, the inspection initialization of the inspected POS machine is completed. If not, the inspection initialization of the inspected POS machine cannot be performed.

3. A POS machine business inspection system according to claim 2, characterized in that: The software and hardware inspection module (2) comprises an operation data acquisition submodule (21), a data analysis submodule (22), a software and hardware information acquisition submodule (23), a path construction submodule (24) and a data acquisition submodule (25); The operation data collection submodule (21) is used to deploy monitoring equipment on the POS machine to be inspected, collect the operation parameters of the POS machine in real time, and obtain the monitoring data of each node of the POS machine, wherein the monitoring data of each node includes hardware monitoring data, software system monitoring data and data security monitoring data; The data analysis submodule (22) is used to input the POS machine monitoring data into a state analysis model to obtain a first analysis result, wherein the first analysis result includes a normal mode and an abnormal mode; The software and hardware information acquisition submodule (23) is used to retrieve the historical inspection information and POS machine location information of a number of POS machines in abnormal modes, wherein the historical inspection information includes a historical inspection time set; A path construction submodule (24), configured to construct an inspection path according to the historical inspection time set and the location information, and generate a first inspection path; A data acquisition submodule (25) is used to perform inspections on the POS machines in the plurality of abnormal modes and acquire first inspection data; The hardware monitoring data includes whether the appearance is intact, whether the buttons are sensitive, whether the printing device can work normally, and whether the connecting cables are loose or damaged; The software system monitoring data includes checking whether the operating system is the latest version, whether there are any uninstalled security patches or system upgrade prompts, and running the built-in diagnostic program to test whether the software functions are normal. The running of the built-in diagnostic program to test whether the software functions are normal includes whether the card swiping is normal, whether the code scanning is normal, and whether the data upload is normal; The data security monitoring data includes verifying whether the data backup mechanism of the POS system is effective, checking the firewall settings, login password strength, and whether the virus protection software status is abnormal.

4. A POS machine business inspection system according to claim 3, characterized in that: The data analysis submodule (22) is used to input the POS machine monitoring data into the state analysis model to obtain the first analysis result. The specific implementation process includes: M1: In the first node of the POS machine to be inspected, collect historical POS machine monitoring data within a preset historical period; M2: Acquire a POS machine monitoring data set, evaluate the historical POS machine monitoring data according to the POS machine monitoring data set, mark the evaluation results that meet the threshold as normal mode, and mark the evaluation results that do not meet the threshold as abnormal mode, and obtain the historical POS machine monitoring data analysis results; M3: annotating the historical POS machine monitoring data analysis results to obtain a first constructed data set; M4: construct a first POS machine state analysis model, and use the first constructed data set to perform deep learning on the first POS machine state analysis model to obtain the first POS machine state analysis model whose accuracy meets a threshold; M5: Continue to construct corresponding status analysis models at other POS machine nodes respectively, and input the monitoring data of each POS machine node into the corresponding device status analysis model to obtain the first analysis results of N nodes.

5. A POS machine business inspection system according to claim 4, characterized in that: The specific implementation process of the path construction submodule (24) includes: N1: According to the location information of several POS machines, obtain the coordinates of several POS machines, where the several coordinates are set as Q; N2: constructing a path for the Q coordinates according to path constraints to generate a path construction result, wherein the path constraints are determined according to the remoteness of the location of a single POS terminal and its usage activity; N3: Obtain the last inspection time and the number of historical inspections of the Q POS terminals according to the historical inspection time set; N4: Prioritize the Q POS terminals according to the last inspection time and the number of historical inspections to obtain the priority of the POS terminals; N5: Adjust the path construction result according to the priority level of the POS machine to obtain a first inspection path.

6. A POS machine service inspection system according to claim 5, characterized in that: The N2 specifically includes: N21: randomly generating an initial first inspection path set according to the path constraint condition and the Q POS machine coordinates, wherein each initial first inspection path represents a type of solution, which is composed of the Q POS machine coordinates; N22: Obtain the distance information between POS terminals, and construct a matching function based on the distance information, as shown in the following formula: Where P represents the matching degree, c represents the distance information between POS machines, n represents the number of pairs of matches among Q POS machines, and i represents any pair of POS machines; N23: Calculate the initial first inspection path set according to the matching function, and generate an optimal path construction result according to the path calculation result.

7. A POS machine service inspection system according to claim 6, characterized in that: The N23 specifically includes: N231: randomly selecting a first inspection path calculation result from the path calculation results as the current optimal calculation result; N232: Continue to randomly select a second inspection path calculation result from the path calculation results; N233: Determine whether the calculation result of the second patrol path is greater than the calculation result of the first patrol path. If so, use the calculation result of the second patrol path as the optimal calculation result; N234: Continue iterating and optimizing until a preset number of iterations is reached, and output the initial first inspection path corresponding to the final optimal calculation result as the optimal path construction result.

8. A POS machine service inspection system according to claim 7, characterized in that: The specific implementation process of the signal inspection module (3) includes: K1: Decompose the signal path of the POS machine to be inspected, determine the signal path set, perform corresponding function analysis on the signal path set, and establish a signal-component relationship list; K2: construct a signal processing unit, which is set in the POS machine and includes a plurality of signal processing sub-units, which correspond to the signal paths; K3: Establishing a transmission relationship between the signal path set and the signal processing unit, connecting each signal path through each signal processing sub-unit, and receiving the signal sent by the POS machine; K4: Based on the signal-component relationship list, set abnormal signal characteristics by collecting historical signal data; K5: Perform feature comparison on the received signals in each signal processing subunit based on the abnormal signal feature, and locate the abnormal components in the POS terminal according to the signal comparison results.

9. A POS machine service inspection system according to claim 8, characterized in that: In the K1, the signal path of the POS terminal to be inspected is decomposed to determine the signal path set, including: K11: Perform signal path analysis on the POS machine to be inspected and determine the signal path devices, wherein the signal path devices include wireless network devices and IoT sensor devices; K12: Based on the signal path information, perform signal path feature analysis to determine the range and connection mode of each signal path; K13: Analyze the signal path characteristics of the POS machine to be inspected to determine the signal path type and signal path range; K14: performing credibility matching based on each signal path range and connection mode, signal path type and signal path range, to obtain the signal path set, where the signal path set is a signal path that satisfies a matching relationship; The K4 specifically includes: K41: Determine the signal type of each component based on the signal-component relationship list; K42: Based on the signal type of each component, collect historical signal data of the component and determine an abnormal signal set; K43: Use deep learning model to identify abnormal signal sets and identify abnormal signal features; The specific implementation process of K43 is as follows: K431: Collect 100,000 pieces of historical signal data of components, where the historical signal data of components are divided into an abnormal signal set and a non-abnormal signal set; K432: Divide 100,000 records into training set, test set and validation set in a ratio of 8:1:1; the data in the training set contains all abnormal signal set information, the data in the test set contains both abnormal signal set and non-abnormal signal set information, and the data in the validation set contains both abnormal signal set information and non-abnormal signal set information; K433: Use annotation tools to annotate the data in the training set and generate an XML file; K434: After extracting features using a convolutional neural network algorithm, the deep learning model trains the data in the training set one by one; K435: The deep learning model recognizes the data in the test set one by one and obtains the accuracy and error rate; K436: If the accuracy and error rate do not meet the requirements, proceed to step K437; if the accuracy and error rate meet the requirements, the deep learning model recognizes the data in the validation set one by one, and if the accuracy and error rate meet the requirements, the training ends; if the accuracy and error rate do not meet the requirements, proceed to step K437; K437: After manually adjusting the recognition weight coefficient of the deep learning model, the data in the validation set are recognized one by one. If the recognition accuracy and error rate meet the requirements, the training ends. If not, the recognition weight coefficient is adjusted again and the data in the validation set are recognized one by one again. After each recognition of the data in the validation set, the recognition record is deleted until the recognition accuracy and error rate meet the requirements and the training ends. K438: The trained deep learning model is used to identify abnormal signal features. If the input signal is found to be an abnormal feature, the signal is set as an abnormal feature. The K5 specifically includes: K51: extract the abnormal signal feature of the signal processing subunit, and if the abnormal signal feature is a strength feature, obtain the abnormal strength interval; K52: Acquire a receiving signal path of the signal processing subunit, and construct a signal strength waveform based on the receiving signal path; K53: Adapt the signal strength waveform to the abnormal coordinates constructed by the abnormal strength interval, obtain the abnormal trigger point, locate the position of the abnormal components in the POS machine, and generate risk feature information.

10. A POS machine service inspection system according to claim 9, characterized in that: The specific implementation process of the risk sharing decision module (4) is as follows: T1: Train the preset risk treatment model using the historical risk treatment data set with risk treatment measure labels added to obtain a corresponding set of risk treatment model training losses; T2: setting the optimal risk processing model training loss as the loss parameter of the preset risk processing model to obtain a pre-trained risk processing model after the risk processing model training; T3: inputting the risk characteristic information into the pre-trained risk processing model to obtain the risk processing measures output by the pre-trained risk processing model; T4: According to the preset risk treatment measure scoring rules and the risk treatment measures, determine the score of the risk treatment measures and obtain a corresponding risk decision report.

Images