On-site service personnel behavior data analysis method, system and device, and medium

Abstract

The invention discloses a field service personnel behavior data analysis method, system and device and a medium. The method comprises the steps of collecting original data of a target scene in real time, and obtaining a field service state according to the original data; when the on-site service state meets a first preset condition, on-site service data in a preset period are obtained, and the on-site service data in the preset period at least comprise on-site service action information, on-site image information and a user electronic signature; when the field service state does not meet the first preset condition, triggering and executing a preset alarm strategy to obtain an execution result; when the execution result meets a second preset condition, triggering a field side acquisition system to obtain field service data in a preset period; and when the execution result does not meet the second preset condition, triggering and executing the alternative strategy, and triggering the field side acquisition system to obtain the field service data in the preset period. According to the method, reliable and efficient field service management and control can be realized.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of computer, and particularly relates to a field service personnel behavior data analysis method and system, device and medium. BACKGROUND

[0002] The existing field service system only realizes "work order level" but not "process level" management, resulting in that the service process forms a data black box in the physical field, and an enterprise cannot collect and quantitatively control five types of variables, i.e., service personnel, service site, tool and material, operation specification and charging standard, which are dispersed in different places, so that the one-time acceptance qualified rate is less than 95% for a long time, and the customer satisfaction is less than 80%.

[0003] The traditional 400 / 800 hotline and CRM order dispatch mode relies on manual agents to complete order building, order dispatch, order urging and follow-up, and the upper limit is reached when the daily processing number of each agent is 30-50, which cannot be linearly expanded with the business scale; and unstructured data such as voice, paper signature and WeChat photo cannot be automatically associated with the work order, and the quality traceability and correction prevention system is missing, resulting in high costs of repeated visits, rework and complaints.

[0004] The existing end-to-end after-sales platform integrates GPS trajectory and electronic work order, but lacks multi-dimensional variable analysis, and cannot predict resource gaps before visiting, check operation compliance after arrival, and automatically check charges and materials before completion. SUMMARY

[0005] Therefore, the embodiments of the present disclosure provide a field service personnel behavior data analysis method and system, device and medium, which can solve the problems that the analysis of the service behavior in the prior art is not comprehensive, the analysis result is unreliable, and reliable and efficient field service management and control cannot be realized.

[0006] In a first aspect, the embodiments of the present disclosure provide a field service personnel behavior data analysis method, comprising: collecting original data of a target scene in real time, wherein the original data comprises personnel image information, personnel path information and target personnel information; obtaining a field service state according to the original data; when the field service state meets a first preset condition, obtaining field service data in a preset period, wherein the field service data in the preset period at least comprises field service action information, field image information and user electronic signature; when the field service state does not meet the first preset condition, triggering and executing a preset alarm strategy, and obtaining an execution result; when the execution result meets a second preset condition, triggering a field side collection system to obtain the field service data in the preset period; When the execution result does not satisfy the second preset condition, an alternative strategy is triggered and executed, and the on-site acquisition system acquires the on-site service data in a preset period.

[0007] In a second aspect, the embodiments of the present disclosure further provide a field service personnel behavior data analysis system, comprising: An acquisition module is configured to acquire original data of a target scene in real time, wherein the original data comprises personnel image information, personnel path information, and target personnel information. A state acquisition module is configured to acquire a field service state according to the original data. A first analysis module is configured to acquire field service data in a preset period when the field service state satisfies a first preset condition, wherein the field service data in the preset period at least comprises field service action information, field image information, and user electronic signature. A second analysis module is configured to trigger a preset alarm strategy and execute the preset alarm strategy when the field service state does not satisfy the first preset condition, and acquire an execution result. A third analysis module is configured to trigger an on-site acquisition system to acquire the field service data in the preset period when the execution result satisfies a second preset condition. A fourth analysis module is configured to trigger an alternative strategy and execute the alternative strategy when the execution result does not satisfy the second preset condition, and trigger the on-site acquisition system to acquire the field service data in the preset period.

[0008] In a third aspect, the embodiments of the present disclosure further provide a computer device, which adopts the following technical solution: The computer device comprises: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the field service personnel behavior data analysis method described in any one of the above.

[0009] In a fourth aspect, the embodiments of the present disclosure further provide a computer readable storage medium, which stores computer instructions; the computer instructions are used to enable a computer to execute the field service personnel behavior data analysis method described in any one of the above.

[0010] In a fifth aspect, the embodiments of the present disclosure further provide a computer program product, which comprises computer programs / instructions, and the computer programs / instructions are executed by a processor to implement the steps of the method described in any one of the above.

[0011] The field service personnel behavior data analysis method disclosed in the application collects original data of a target scene in real time, obtains a field service state according to the original data, converts the on-site authenticity, which is a subjective fact that cannot be quantified in the past, into a verifiable objective data object, obtains field service data in a preset period when the field service state meets a first preset condition, the field service data in the preset period at least including field service action information, field image information and user electronic signature, triggers a preset alarm strategy and executes when the field service state does not meet the first preset condition, obtains an execution result, triggers a field side acquisition system to obtain field service data in a preset period when the execution result meets a second preset condition, triggers an alternative strategy and executes when the execution result does not meet the second preset condition, and triggers the field side acquisition system to obtain field service data in a preset period, changes the risk in the service performance process from a post-facto passive compensation mode to an in-process active intervention mode, realizes reliable and efficient field service management and control, that is, the application can realize millisecond-level acquisition, second-level decision and chain-level evidence storage in the whole life cycle of service personnel behavior data through an end-edge-chain collaborative architecture.

[0012] The above description is only a summary of the technical solutions of the present disclosure, in order to more clearly understand the technical means of the present disclosure, the content of the specification can be implemented, and in order for the above and other purposes, features and advantages of the present disclosure to be more obvious and easy to understand, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present disclosure, and those skilled in the art can also obtain other drawings according to these drawings without creating any creative labor.

[0014] Figure 1 The flowchart of the field service personnel behavior data analysis method provided by the embodiments of the present disclosure is shown.

[0015] Figure 2 The flowchart of the method for collecting original data of a target scene in real time provided by the embodiments of the present disclosure is shown.

[0016] Figure 3 The flowchart of the method for obtaining a field service state according to original data provided by the embodiments of the present disclosure is shown.

[0017] Figure 4 The flowchart of the method for obtaining field service data in a preset period provided by the embodiments of the present disclosure is shown.

[0018] Figure 5A structural schematic diagram of a computer device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0019] The embodiments of the present disclosure will be described in detail below with reference to the drawings.

[0020] It should be apparent that the following describes embodiments of this disclosure by way of specific examples, and that one skilled in the art could readily derive further advantages and effects from this disclosure. Obviously, the described embodiments are only a part of the embodiments of this disclosure, rather than all the embodiments. This disclosure can also be implemented or applied by other different specific embodiments, and various modifications or changes can be made to the details based on different views and applications without departing from the spirit of this disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in this disclosure, all other embodiments obtained by one of ordinary skill in the art without creative labor are within the scope of protection of this disclosure.

[0021] It should be noted that the various aspects of the embodiments described below are within the scope of the appended claims. It should be apparent that the aspects described herein can be embodied in a wide variety of forms and that any specific structure and / or function described herein is merely illustrative. Based on the teachings herein one skilled in the art should appreciate that an aspect described herein can be implemented independently of any other aspects and that two or more of these aspects can be combined in various ways. For example, an apparatus can be implemented or a method can be practiced using any number of the aspects described herein. In addition, such an apparatus can be implemented or such a method can be practiced using other structure and / or functionality in addition to or other than one or more of the aspects described herein.

[0022] It should also be noted that the drawings provided in the following embodiments are only schematic and that the drawings show only those components that are related to the disclosure, whereas other components are not shown so as not to obscure the disclosure with unnecessary detail.

[0023] In addition, in the following description, specific details are provided to facilitate thorough understanding of examples. However, one skilled in the art will understand that the described aspects can be practiced without these specific details.

[0024] Reference Figure 1 The present application discloses a field service personnel behavior data analysis method, comprising: S100, collecting original data of a target scene in real time.

[0025] The original data includes personnel image information, personnel path information, and target personnel information.

[0026] S200, obtaining an on-site service state according to the original data; S300, when the on-site service state meets a first preset condition, obtaining a number of on-site services in a preset period.

[0027] The on-site service data in the preset period includes on-site service action information, on-site image information, and user electronic signatures.

[0028] S400, when the on-site service state does not meet the first preset condition, triggering and executing a preset alarm strategy to obtain an execution result; S500, when the execution result meets a second preset condition, triggering an on-site collection system to obtain on-site service data in a preset period; S600, when the execution result does not meet the second preset condition, triggering and executing an alternative strategy, and triggering the on-site collection system to obtain on-site service data in a preset period.

[0029] Reference Figure 2 For the method of S100, real-time collection of original data of a target scene, the method comprises the following steps: S110, deploying a horizontal federated learning master node in a multi-cloud edge collaboration gateway related to on-site services.

[0030] Specifically, a multi-cloud edge collaboration gateway is newly built between an operator 5G MEC machine room and a party A cloud VPC, and the hardware is an ARM server + TPM security chip; a horizontal federated learning master node program is preinstalled in the gateway, the program is self-equipped with a model splitter, a security aggregator, and a chain light client; the master node exposes a gRPC-over-TLS interface to the outside, and is connected to a 5G core network UPF through SR-IOV in the inside, so that the local server of an outsourcing party can reach the target only by using an internal network slice, without exposing a public IP.

[0031] In this step, the federated hub is sunk to the edge, the delay is less than 15 ms, and the gradient upload failure caused by the jitter of the inter-provincial backbone network is avoided; the TPM chip locks the private key of the master node in the hardware, so that even if the gateway is physically hijacked, the aggregated result cannot be forged, and the requirement of the third level of network security protection for a trusted execution environment is met.

[0032] S120, based on the horizontal federated learning master node, pulling the desensitized historical work orders in the form of gradients from N local servers of outsourcing parties, obtaining a cross-domain comparable behavior baseline weight vector after differential privacy training, and writing the vector into a cache library of an edge collection terminal.

[0033] Specifically, each outsourcing party local server can automatically train the previous day's desensitized work order locally at a preset time every day to output a 128-dimensional gradient vector; the local server calls libsodium to add Laplace noise (ε=0.8) to the gradient, and then packs it into protobuf; after the master node receives N=5 gradients, it performs FedAvg aggregation according to the data volume to obtain the global gradient, and updates once to obtain the behavior baseline weight vector; the master node encrypts the baseline weight vector w0 into Cw0 using the CKKS homomorphic public key, writes it into the alliance chain together with the version number and timestamp, and returns the hash Hchain; the gateway pushes Cw0 to the cache library of the on-site edge collection terminal (industrial box) through MQTT-over-TLS, and the terminal's built-in security chip can only decrypt but cannot export plaintext. Among them, N≥2.

[0034] In this step, the gradient and noise are doubly desensitized, and the party A cannot infer any individual work order from w0. The on-chain hash makes the w0 version non-repudiable, and in the event of a subsequent dispute, it can be used as electronic evidence.

[0035] S130, when the service personnel mobile terminal SDK starts, extract the behavior baseline weight vector corresponding to the on-site from the cache library, denoted as the target vector, and use the target vector as the gating parameter to real-time return the encrypted original data packet containing at least the face feature code, tool information, GPS coordinates, and on-site timestamp.

[0036] Among them, service personnel refers to on-site operating personnel such as visiting engineers, installation / maintenance technicians, etc.; mobile terminal refers to the company App (Android / iOS mobile phone or PDA) in their hands; SDK (Software Development Kit) refers to a piece of development package / function component embedded in the App, rather than a standalone App. Start-up refers to the moment when the engineer opens the App (or the App is awakened by the system), and the initialization code of the SDK begins to execute.

[0037] Specifically, the service personnel hits the mobile device App instantly, the SDK first verifies the chain Hchain to confirm that the version has not been tampered with; the security chip decrypts the plaintext target vector w0 using the preset CKKS private key; the TinyGateNet hardware accelerator splits w0 into 5 sub-vectors, which are respectively mapped to the face, tooling, GPS, clock, and consumable five-way sensor sampling frequencies: if w0 shows a high historical tardiness probability, the GPS sampling frequency is automatically increased to 1 Hz, otherwise it remains at 0.2 Hz; if w0 shows a high tooling misfit probability, the camera frame rate is increased from 15 fps to 30 fps; only the data judged by TinyGateNet as valid frames enters the secure area for SM4 encryption, and is packaged into an encrypted raw data packet (containing a 256-bit face feature code, tooling RFID number, GPS, UTC); the encrypted raw data packet is hashed before being written into the chain again, forming a collection-evidence dual anchor.

[0038] In this step, the gating mechanism can filter out 85% of invalid frames, saving about 30% of 4G / 5G traffic fees per month, while reducing the decoding pressure of the cloud GPU; the face feature code is extracted as a vector in the chip, and the double hashing on the chain makes it impossible for a single party to tamper with "which second, which person, which picture" afterwards.

[0039] Reference Figure 3 For the method of S200 "obtaining the on-site service state according to the raw data", it includes: S210, parsing the encrypted raw data packet, extracting the GPS-time stamp sequence, calculating the path deviation of the preset path and the actual path, if the path deviation is not greater than the preset threshold, outputting a first state code, otherwise outputting a second state code.

[0040] The first state code represents that the person has arrived on time; the second state code represents that the person has not arrived on time.

[0041] Specifically, after the edge industrial control box receives the encrypted raw data packet, it is first decrypted in the security chip using the preset SM4 key to obtain the plaintext GPS-time stamp sequence; the box has a local SQLite that stores a planned polyline, which is derived from the party A work order system and is issued together with the baseline weight vector when the order is dispatched; the GPS point is taken every 30s, and the vertical distance from the point to the planned polyline is calculated to obtain the path deviation; the preset threshold can be flexibly set according to the specific scene; if the path deviation of the last 3 windows is not greater than the preset threshold, output the first state code M1 (the person has arrived on time).

[0042] If any window path deviation is greater than a preset threshold, a second status code M0 (person not timely arrived) is immediately output, and the first time of crossing the boundary is recorded as t0; the status code is accompanied by a 128-bit hash fingerprint, which covers the path deviation, the preset threshold, the number of crossing points, and UTC, to prevent subsequent tampering by a single party.

[0043] S220, write the first status code or the second status code and the encrypted original data packet into a message queue.

[0044] Specifically, the edge box packs the status code + encrypted original data packet + hash fingerprint into a message body, writes it into a local RocksDB message queue using MQTT-over-TLS 1.3, and the queue topic is sharded according to the work order number-year-month; after successfully writing the database, immediately calculate the SHA-256 of the whole message, and write the digest to the alliance chain again to form a queue-chain double anchor; only after the write-chain transaction returns an acknowledgement, the box uploads the message body to the A party cloud through a 5G slice, and the cloud consumer group consumes it in order according to the partition, ensuring that the “first judge state, then trigger alarm / collection” timing is strictly consistent; if the network is disconnected, the message is persisted in RocksDB, and after the network is restored, it is automatically de-duplicated to avoid repeated alarms.

[0045] Reference Figure 4 For the method of S300 “when the on-site service state meets the first preset condition, obtaining on-site service data in a preset period”, that is, the method of obtaining on-site service data in a preset period, specifically includes: S310, when receiving the first status code, the edge side immediately starts a lightweight CNN-Transformer double-branch model to perform instance segmentation on each frame of image uploaded on site to generate a 128-dimensional behavior feature vector.

[0046] Specifically, after receiving the first status code, the edge industrial computer box triggers a quality spot check timer, and sets the subsequent 30 minutes as the preset period by default; the box is embedded with an 8 TOPS NPU, and a lightweight CNN-Transformer double-branch model is preset, the CNN branch is responsible for spatial details (safety helmet, gloves, tool position), and the Transformer branch is responsible for temporal relationship (pick up->put->tighten->insert four-step sequence); the model performs instance segmentation on each frame of 720p image uploaded on site at 6fps, outputs a 128-dimensional behavior feature vector, and the single-frame delay is 38ms. The behavior feature vector is calculated and walked in the chip, the original image is automatically discarded, only the behavior feature vector and the corresponding UTC timestamp are retained, which meets the privacy requirement of not exporting images but only exporting vectors; it is packed once every 1s (6 frames) and written into a local circular buffer for real-time access by S320.

[0047] In this step, edge inference can save 95% of upstream bandwidth (only 128-dimensional vectors are transmitted, not 3MB original pictures); CNN-Transformer dual branch can improve 12% of action sequence recognition rate compared with pure CNN, and reduce the timing error such as wearing gloves first and then picking up tools; Chip-level calculation and picture deletion strategy make the employer not need to build expensive face desensitization server, and check through supervision on site.

[0048] S320, the behavior feature vector and the standard template vector in the SOP knowledge graph are calculated by cosine similarity, and the similarity score is obtained; if the similarity scores of the continuous k frames are all less than the preset similarity threshold, the score items are written back to the SLA score in real time, and a low similarity picture set is generated synchronously.

[0049] Specifically, the edge box is pre-installed with a lightweight SOP knowledge graph, and the graph nodes are standard action template vectors, which are trained in advance by the employer's process department in a federal way and are issued together with the behavior baseline weight vector; the cosine similarity of each row behavior feature vector and the corresponding standard action template vector is calculated, and the threshold is dynamically taken from the worker-difficulty sub-vector in the behavior baseline weight vector: for example, the threshold of high-altitude electrician is 0.82, and the threshold of indoor installation and maintenance is 0.75.

[0050] If the cosine similarity of the continuous k frames is not greater than the threshold value, the following is triggered immediately: 1) write the score items-2 points to the SLA score, and take UTC, work order number, and similarity average; 2) pack the 128-dimensional behavior feature vector, similarity sequence, and UTC of the k frames into a low similarity picture set summary (still without original picture), calculate SHA-256, and write to the alliance chain; if there is no continuous k frames that do not meet the standard within a preset period (such as 30s) thereafter, the score recorder automatically rebounds +1 point, forming a reversible deduction mechanism to avoid excessive punishment caused by accidental occlusion.

[0051] For the method of S400 "when the on-site service state does not meet the first preset condition, triggering and executing a preset alarm strategy, and obtaining an execution result", specifically comprising: S410, when receiving the second state code, calling the AI voice telephone engine, taking the service personnel number, delay minutes, and on-chain historical SLA score as TTS variables, and automatically generating a personalized reminder voice.

[0052] Specifically, the edge box pushes M0 (containing the service personnel number, the first time crossing the border time t0, and the current UTC) to the operator 5G message gateway, and the gateway immediately transfers the delay alarm event to the cloud AI voice telephone engine; the engine first reads the service personnel's past preset day SLA score S0 (full score 100, chain checkable and unchangeable) from the alliance chain, and calculates the delay minutes ΔT = now-t0. If S0≥90, the script is "Respectful gold medal engineer XXX, the system shows that you are ΔT minutes late than the planned time, do you still go as scheduled?". If 70≤S0<90, the script adds a reminder "Please note that you have been late twice this month, and further delay will affect the star rating". If S0<70, the script adds "This time, if you are late again, you will trigger a deduction of 50 yuan, do you confirm to continue to go". The outbound number uses the enterprise unified customer service number, showing the brand after-sales, avoiding the service personnel from rejecting the call due to unfamiliar numbers; the whole conversation is recorded, and the recording hash is written back to the alliance chain within 5 seconds after the call ends, forming the on-chain voice evidence.

[0053] S420, executing the personalized reminder voice and obtaining user key feedback results; Encode the user key feedback results as execution results and write them into the message queue.

[0054] The execution result is a first value or a second value. In this embodiment, the first value is preferably 1, and the second value is preferably 0. When the execution result is 1, it means that the person notified by the voice call agrees to go; when the execution result is 0, it means that the person notified by the voice call cannot go.

[0055] Specifically, at the moment of hanging up the call, the IVR parses the key results as: key 1-execution result R0=1 (agree to go), key 0-execution result R0=0 (cannot go), timeout no key-considered as 0, and R0 is packaged into JSON: {E-i, order number, R0, UTC, recording hash}; write to topic Q2 using MQTT-over-TLS 1.3, and the message body is calculated again SHA-256 to write the alliance chain, forming the response-locking double anchor; if R0=1, the subsequent order dispatching system continues the original order; if R0=0, immediately trigger the reassignment or standby engineer process, and pre-deduct 50 yuan of breach of contract fine from the employee. Complete key→coding→chain within 3 seconds, ensure that the same second, same hash, and unalterable snapshot are always obtained by the subsequent reassignment / deduction module; 1 / 0 simple coding allows any downstream system to directly use if-else without parsing complex semantics, reducing integration costs; on-chain double anchoring (recording hash+response hash) allows insurance claims, monthly settlement, and labor arbitration to share the same evidence, reducing 90% of manual reconciliation work hours.

[0056] The first preset condition is satisfied when the target personnel has arrived at the scene on time; and the first preset condition is not satisfied when the target personnel has not arrived at the scene on time.

[0057] For the method of S500 "when the execution result meets the second preset condition, the on-site side acquisition system acquires the on-site service data in a preset period", specifically comprising: S510, when the execution result is the first numerical value, the high-definition photographing permission is unlocked, and the service personnel are forced to upload at least the on-site photos of tools, materials and gestures within a preset time.

[0058] Specifically, after the edge box receives R0=1, it immediately pushes a photographing unlocking instruction to the service personnel's mobile phone terminal, and at the same time, changes the work order status light from "gray" to "blue", indicating "promised, waiting for self-evidence"; the mobile phone terminal SDK forces to open the 4K camera, and the interface is locked to the three-in-one shooting template: a) tool in the same frame: multimeter, insulating rod and safety belt must be in the same frame; b) material in the same frame: cable, connector and label must be clear and identifiable; c) gesture in the same frame: left hand pinching label, right hand holding tool, forming double-hand occupation evidence to prevent walking away after posing; the system presets a 10 min countdown, and if it is not transmitted within the time, it will automatically write back a photographing missing warning and deduct 5 points; the photo is uploaded through the 5G slice channel, and the edge side only does pixel-level hashing without saving the original picture, meeting the non-picture output compliance requirements.

[0059] In this step, the three-in-one shooting template locks the person, machine and material in one photo at a time, and if a dispute occurs later, the on-site can be restored directly using the on-chain hash; the 10 min forced window seamlessly connects "promise to" and "real on-site", eliminating the vacuum period of "saying to go in the phone, but not actually going"; the 4K original picture is only cached on the mobile phone terminal for 30 min, and the edge only stores the hash, saving 90% of edge storage.

[0060] S520, send the on-site photos to the CNN-Transformer double-branch model, output the updated behavior judgment threshold, and synchronize back to the edge cache.

[0061] Specifically, the edge NPU sends the photo into the CNN-Transformer dual-branch model, the CNN branch extracts tool-material space features, and the Transformer branch extracts gesture sequential timing features; the model outputs a 128-dimensional environment feature vector Ve, and the cosine similarity with the SOP standard template is obtained to obtain the environmental offset degree AS; use AS to dynamically refresh the behavior judgment threshold of the next 30 minutes: if AS >= 0.9 (light, tool, background and training library are highly consistent), the threshold remains default tau0; if 0.7 <= AS < 0.9 (overcast, reflection, narrow space), the threshold is lowered by 5%, reducing false negatives; if AS < 0.7 (high altitude, night, strong backlight), the threshold is lowered by 10%, and the "safety helmet not worn" score is reduced from -5 to -3, avoiding excessive punishment caused by environmental interference; the new threshold tau' is written to the edge cache, and the UTC and Ve hash are written together to form a threshold update voucher for end-of-month audit of why the same action is not deducted this time.

[0062] In this step, one photo can make the model understand the on-site environment, and the real-time error rate in the next 30 minutes can be reduced from 12% to 3%; the threshold is recorded on the chain to prevent artificial water from being put in by shooting in the head, and also to prevent one-size-fits-all mistakes, so that the threshold is lowered without lowering the standards; Ve hash corresponds to photo hash one by one, and if it is found that the threshold has been maliciously raised, Ve can be restored directly using the on-chain hash, and responsibility tracing is completed.

[0063] Among them, meeting the second preset condition means that the target service personnel promises to arrive at the scene according to the agreed time when notified by voice call, although they do not arrive at the scene within the specified time, wherein the system preset specified time refers to the time in advance relative to the agreed time, the purpose is to monitor the advance rate of the target personnel, and to ensure the on-site arrival rate.

[0064] For the method of S600 "when the execution result does not meet the second preset condition (i.e. the target service personnel cannot arrive due to business), triggering the alternative strategy and executing, and triggering the on-site acquisition system to acquire on-site service data within a preset period", specifically includes: S610, when the execution result is the second value, the smart contract reads the real-time SLA ranking on the chain, freezes and transfers the preset percentage of the original order allocation quota to the first ranked alternative outsourcing party, and generates a new order allocation matrix.

[0065] Specifically, the edge box pushes the R0=0 event to the smart contract, which immediately queries the latest 7-day SLA real-time ranking on the chain; locks the original service personnel E-old's current 10% order allocation quota (e.g., 30 orders / month), denoted as quota Q; transfers Q to the Top-1 ranked alternative outsourcing party E-new, and generates a freeze-transfer transaction Tx on the chain, which contains: original order number, E-old's frozen quota Q, E-new's obtained quota Q, and transfer reason code "R0=0".

[0066] The contract synchronously refreshes the order allocation matrix D→D', in which E-old's available orders decrease by Q and E-new's available orders increase by Q, and a UTC timestamp is added; the transaction Tx hash is immediately written back to the dispatch center, and the downstream TMS (transport management system) reorders according to D' within 30 seconds.

[0067] In this step, on-chain second-level freezing prevents manual fraud, and the original engineer cannot regain the quota through private negotiations; the 10% proportion both punishes the latecomers and does not excessively harm their monthly income, reducing the risk of labor arbitration; the Top-1 alternative directly benefits, encouraging all outsourcing parties to regard SLA as a lifeline, and the ranking is refreshed every 5 minutes, forming a reliable mechanism.

[0068] S620, the freeze-transfer event and the new order allocation matrix are returned to the data lake, and a four-dimensional attribution heat map including behavior, quality, timeliness, and cost is generated using SHAP values.

[0069] Specifically, the data lake can subscribe to Tx events and pull 200+ features of the last 90 days in four categories: behavior, quality, timeliness, and cost; use the trained credit redistribution XGBoost model to predict "if Q orders are transferred to E-new, how much is the expected customer cancellation rate reduced", output the predicted value ΔC; SHAP values calculate the contribution of each feature to ΔC, and generate a four-dimensional attribution heat map R: 1) behavior dimension: E-old's R0=0 times in the last 30 days→contribution +32%; 2) quality dimension: E-new's one-time repair rate 98%→contribution +28%; 3) timeliness dimension: E-new's average door-to-door time 18 minutes→contribution +25%; 4) cost dimension: E-new's single-kilometer cost +0.8 yuan→contribution -15%.

[0070] The R graph can be displayed with "red-blue" gradient, with red area as negative factors and blue area as positive factors, and with "explainable text": "because E-old has failed to arrive 3 times this month, the customer cancellation risk increases by 3.2%; selecting E-new can reduce it by 2.8%, with a net benefit of 2.3%." The R graph is written back to the alliance chain together with Tx and D', for financial, outsourcing, and regulatory personnel to scan and view.

[0071] In this step, the four-dimensional attribution makes it clear why to freeze him and why to choose it, and the outsourcing party no longer questions the black box operation, which can effectively reduce the number of complaints; The heat map can directly point out the percentage of cost increase, and the finance can fine-tune the unit price in advance at the settlement end to avoid after-the-fact wrangling; The same R chart can be used by the regulatory side to select the annual high-quality outsourcing, reduce repeated audits, and effectively save the review manpower.

[0072] The field service personnel behavior data analysis method disclosed in the application, real-time acquisition of original data of a target scene, obtaining a field service state according to the original data, converting the on-site authenticity, a subjective fact that cannot be quantified in the past, into a verifiable objective data object; when the field service state meets a first preset condition, obtaining field service data in a preset period, the field service data in the preset period at least including field service action information, field image information and user electronic signature; when the field service state does not meet the first preset condition, triggering a preset alarm strategy and executing, obtaining an execution result; when the execution result meets a second preset condition, triggering the field side acquisition system to obtain the field service data in the preset period; when the execution result does not meet the second preset condition, triggering an alternative strategy and executing, and triggering the field side acquisition system to obtain the field service data in the preset period, changing the risk in the service performance process from a post-facto passive compensation mode to an in-process active intervention mode, realizing reliable and efficient field service control, reducing quality default penalties, reducing customer service representative costs, improving user net recommendation values, generating financial, insurable and cross-platform mutually recognized performance data assets, and realizing synchronous gains in technical, commercial and compliance effects.

[0073] Through the end-edge-chain collaborative architecture, the application can realize millisecond-level acquisition, second-level decision and chain-level evidence storage in the whole life cycle of service personnel behavior data. Specifically, a path-image-time sequence three-module fusion model is used to jointly detect service personnel GPS drift, stay duration and face disappearance events, and AI voice intervention is triggered within a preset time before the occurrence of a late arrival. Compared with traditional post-alarm, the real late arrival rate is effectively reduced.

[0074] The three-in-one technical solution of the process level, the chain level and the explainable level proposed in the application can effectively solve the three problems of the existing technology, i.e., the work order level black box, the artificial ceiling and the multi-dimensional variable unaccounting. Specifically, the existing technology can only record the creation / closure of the work order. The application divides one service into ≥15 standard processes through the five-dimensional collection module of edge side image-path-action-material-charge. The CNN-Transformer hybrid model is triggered for each process to output 0 / 1 pass signals in real time and write them into the chain. The existing CRM needs to compare the map, skills and inventory by the agent manually. The application deploys an XGBoost-LSTM hybrid model offline to predict the four-dimensional gap of man-place-material-charge in the future with a granularity of 5 minutes and automatically generate a gap vector readable by the smart contract on the chain. Traditional voice, paper signature and WeChat photos cannot be automatically bound with the work order ID. The application generates a three-dimensional hash Htx of the five-tuple of user electronic signature+GPS+UTC+material code+charge code through the national encryption SM2 at the moment of closing the order and writes it into the alliance chain to realize the professional technical conversion of unstructured-structured-chain structured. The existing platform lacks quantitative means of economic punishment for abnormalities such as late delivery, overcharging and inconsistent materials. The application uses SHAP values to perform four-dimensional attribution of freezing-transfer events on behavior-quality-time-cost, generates a heat map and uploads it to the Tx to realize the three synchronization of decision-making-punishment-explanation.

[0075] The field service personnel behavior data analysis method disclosed in the application can realize effective management and control of SPM field service, real-time monitoring and preventive management of the field service process by an intelligent system, and control of each key operation of the field service personnel by service standard specification SOP. The field service record is entirely managed by visual data (action time point, position coordinates, field photos and videos, user electronic signature, etc.), which can not only greatly improve the quality management level and efficiency of enterprise field service business, but also greatly reduce labor costs.

[0076] In a second aspect, the application discloses a field service personnel behavior data analysis system for executing the field service personnel behavior data analysis method disclosed in the first aspect of the application. The system comprises: A collection module is configured to collect raw data of a target scene in real time. The raw data includes personnel image information, personnel path information and target personnel information. A state acquisition module is configured to obtain a field service state according to the raw data. A first analysis module is configured to acquire field service data in a preset period when the field service state meets a first preset condition. The field service data in the preset period includes at least field service action information, field image information and user electronic signature. The second analysis module is configured to trigger a preset alarm strategy and execute when the on-site service state does not satisfy the first preset condition, and obtain an execution result. The third analysis module is configured to trigger the on-site side acquisition system to acquire on-site service data in a preset period when the execution result satisfies a second preset condition. The fourth analysis module is configured to trigger an alternative strategy and execute when the execution result does not satisfy the second preset condition, and trigger the on-site side acquisition system to acquire on-site service data in a preset period.

[0077] A computer device according to an embodiment of the present disclosure includes a memory and a processor. The memory is configured to store non-transitory computer readable instructions. Specifically, the memory can include one or more computer program products, which can include various forms of computer readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM), cache memory, and / or the like. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, and / or the like.

[0078] The processor can be a central processing unit (CPU) or other form of processing unit having data processing and / or instruction execution capabilities, and can control other components in the computer device to perform desired functions. In an embodiment of the present disclosure, the processor is configured to run the computer readable instructions stored in the memory, so that the computer device performs all or part of the steps of the on-site service personnel behavior data analysis method according to the embodiments of the present disclosure.

[0079] Those skilled in the art will understand that, in order to solve the technical problem of how to obtain a good user experience effect, the present embodiment can also include well-known structures such as a communication bus, an interface, and the like, which should also be included in the protection scope of the present disclosure.

[0080] As Figure 5 A structural schematic diagram of a computer device according to an embodiment of the present disclosure is provided. It shows a structural schematic diagram of a computer device suitable for use to implement the computer device in the embodiments of the present disclosure. Figure 5 The computer device shown is merely an example, and should not impose any limitation on the functions and use range of the embodiments of the present disclosure.

[0081] As Figure 5As shown, the computer device can include a processor (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) or loaded from a storage device into a random access memory (RAM). Various programs and data required for the operation of the computer device are also stored in the RAM. The processor, the ROM, and the RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0082] Generally, the following devices can be connected to the I / O interface: input devices including, for example, sensors or visual information collection devices; output devices including, for example, display screens; storage devices including, for example, magnetic tapes, hard disks, etc.; and communication devices. The communication devices can allow the computer device to communicate wirelessly or wired with other devices, such as edge computing devices, to exchange data. Although Figure 5 The computer device is shown with various devices, but it should be understood that not all of the shown devices are required to be implemented or present. More or fewer devices can alternatively be implemented or present.

[0083] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program according to embodiments of the present disclosure. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication device, or installed from the storage device, or installed from the ROM. When the computer program is executed by the processor, all or part of the steps of the field service personnel behavior data analysis method of embodiments of the present disclosure are performed.

[0084] Detailed descriptions of the embodiments can refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0085] The computer-readable storage medium according to embodiments of the present disclosure has non-transitory computer-readable instructions stored thereon. When the non-transitory computer-readable instructions are run by a processor, all or part of the steps of the field service personnel behavior data analysis method of the foregoing embodiments of the present disclosure are performed.

[0086] The computer-readable storage medium described above includes, but is not limited to, optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or a mobile hard disk), media with built-in rewritable non-volatile memory (e.g., a memory card), and media with built-in ROM (e.g., a ROM cartridge).

[0087] The detailed description of the embodiments hereinabove with reference to the drawings are applicable to this example. Repetition is omitted here.

[0088] The above describes the basic principles of the present disclosure in combination with specific examples, but it should be noted that the advantages, benefits, effects and the like mentioned in the present disclosure are only examples and are not limiting, and these advantages, benefits, effects and the like cannot be considered as necessary for each embodiment of the present disclosure. In addition, the above specific details of the disclosure are only for the purpose of example and for the purpose of understanding, and are not limiting, and the above details do not limit the disclosure to be necessarily implemented with the above specific details.

[0089] In the present disclosure, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. The block diagrams of the devices, apparatuses, equipment, systems involved in the present disclosure are only illustrative examples and are not intended to require or imply the connection, arrangement, configuration as shown in the block diagram. As those skilled in the art will recognize, these devices, apparatuses, equipment, systems can be connected, arranged, configured in any manner. Words such as "include", "contain", "have" and the like are open-ended words, which mean "including but not limited to", and can be used interchangeably. The words "or" and "and" used herein mean the word "and / or", and can be used interchangeably unless the context clearly indicates otherwise. The word "such as" used herein means the phrase "such as but not limited to", and can be used interchangeably.

[0090] In addition, as used herein, "or" used in a list of items, prefaced by "at least one of", indicates a disjunctive list such that, e.g., a list of "at least one of A, B, or C" means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). In addition, the phrase "example of" does not mean an example preferred or better than other examples.

[0091] It should also be noted that in the systems and methods of the present disclosure, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalents of the present disclosure.

[0092] Various changes, modifications, and alterations to the techniques described herein can be made without departing from the teachings of the attached claims. Moreover, the scope of the claims of this disclosure is not limited to the particular aspects described above. In addition, where a process, machine, manufacture, composition of matter, means, method, or result containing procedural, business, and other steps is described, it is understood that the description is meant to encompass the specific implementation of the steps described, as well as the substitution of equivalent steps, or equivalent steps in the performance order. Accordingly, the attached claims are to be interpreted as embracing the specific aspects and embodiments described herein, as well as future modifications, changes, and alterations of the aspects and embodiments.

[0093] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other aspects without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0094] The above description has been presented for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the disclosure to the forms disclosed herein. Although several example aspects and embodiments have been discussed above, those of ordinary skill in the art will appreciate a variety of modifications, alternatives, permutations, additions, and sub-combinations of the aspects and embodiments disclosed herein.

Claims

1. A field service personnel behavior data analysis method, characterized by, The method comprises the following steps: Real-time acquisition of original data of a target scene, the original data comprising personnel image information, personnel path information, and target personnel information; Obtaining a field service state according to the original data; When the field service state meets a first preset condition, acquiring field service data in a preset period, the field service data in the preset period comprising at least field service action information, field image information, and user electronic signature; When the field service state does not meet the first preset condition, triggering a preset alarm strategy and executing the same to obtain an execution result; When the execution result meets a second preset condition, triggering a field-side acquisition system to acquire field service data in a preset period; When the execution result does not meet the second preset condition, triggering an alternative strategy and executing the same, and triggering the field-side acquisition system to acquire field service data in a preset period.

2. The field service personnel behavior data analysis method of claim 1, wherein, The real-time acquisition of original data of a target scene comprises the following steps: Deploying a horizontal federated learning master node in a multi-cloud edge collaboration gateway related to field service; Pulling gradient-form desensitized historical work orders from N outsourced local servers based on the horizontal federated learning master node, obtaining a cross-domain comparable behavior baseline weight vector after differential privacy training, and writing the vector into a cache library of an edge acquisition terminal; When a service personnel mobile terminal SDK is started, extracting a behavior baseline weight vector corresponding to a field from the cache library, denoted as a target vector, and using the target vector as a gating parameter to real-time return an encrypted original data packet containing at least face feature code, tool information, GPS coordinates, and field timestamp.

3. The field service personnel behavior data analysis method of claim 2, wherein, The obtaining of a field service state according to the original data comprises the following steps: Extracting a GPS-time stamp sequence from the encrypted original data packet, calculating a path deviation between a preset path and an actual path, and if the path deviation is not greater than a preset threshold, outputting a first state code, otherwise outputting a second state code; Writing the first state code or the second state code and the encrypted original data packet into a message queue.

4. The field service personnel behavior data analysis method of claim 3, wherein, The acquisition of field service data in a preset period when the field service state meets a first preset condition comprises the following steps: When the first state code is received, starting a CNN-Transformer double-branch model, performing instance segmentation on each frame of image uploaded in the field to generate a 128-dimensional behavior feature vector; Calculating a cosine similarity between the behavior feature vector and a standard template vector in a SOP knowledge graph to obtain a similarity score; If the similarity scores of k consecutive frames are all less than a preset similarity threshold, real-time writing a deduction item to a score recorder.

5. The field service personnel behavior data analysis method of claim 4, wherein, The triggering of a field-side acquisition system to acquire field service data in a preset period when the execution result meets a second preset condition comprises the following steps: When the execution result is a first numerical value, unlocking a high-definition photographing permission, and notifying a service personnel to upload field photos containing at least tools, materials, and gestures within a preset time; Sending the field photos to the CNN-Transformer double-branch model, outputting an updated behavior judgment threshold, and synchronously writing the threshold to an edge cache.

6. The field service personnel behavior data analysis method of claim 5, wherein, The execution result does not meet the second preset condition, triggers the alternative strategy and executes, and triggers the field side acquisition system to acquire the field service data in a preset period, including: When the execution result is a second value, the smart contract reads the real-time SLA ranking on the chain, freezes and transfers a preset percentage of the original order allocation quota to the first ranked alternative outsourcing party, generates a new order allocation matrix; The current freezing-transfer event and the new order allocation matrix are backflowed to the data lake, and a four-dimensional attribution heat map including behavior, quality, timeliness and cost is generated using SHAP values.

7. A field service personnel behavior data analysis system characterized by, It includes: The acquisition module is used for real-time acquisition of original data of a target scene, and the original data includes personnel image information, personnel path information and target personnel information; The state acquisition module is used for obtaining a field service state according to the original data; The first analysis module is used for acquiring field service data in a preset period when the field service state meets a first preset condition, and the field service data in the preset period at least includes field service action information, field image information and user electronic signature; The second analysis module is used for triggering a preset alarm strategy and executing when the field service state does not meet the first preset condition, and obtaining an execution result; The third analysis module is used for triggering the field side acquisition system to acquire the field service data in the preset period when the execution result meets a second preset condition; The fourth analysis module is used for triggering the alternative strategy and executing when the execution result does not meet the second preset condition, and triggering the field side acquisition system to acquire the field service data in the preset period.

8. A computer apparatus, comprising: The computer device includes: At least one processor; and The memory is in communication connection with the at least one processor; wherein The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the field service personnel behavior data analysis method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions; the computer instructions are used for making the computer execute the field service personnel behavior data analysis method of any one of claims 1-6.

10. A computer program product comprising computer instructions, characterized in that, The computer instructions are executed by the processor to realize the steps of the method of any one of claims 1-6.

Images