Inspection service management system based on artificial intelligence and internet of things

Abstract

The application relates to the technical field of artificial intelligence and Internet of Things, in particular to a patrol inspection service management system based on artificial intelligence and Internet of Things, which comprises a data acquisition module, an edge computing module, a task allocation module, a path planning module, a prewarning module and a user interaction module; wherein: the data acquisition module is used for collecting multi-dimensional data of an environment and equipment in real time; the edge computing module is used for preliminarily analyzing and processing the multi-dimensional data; the task allocation module is used for generating a patrol inspection task according to a patrol inspection rule; the path planning module is used for generating an optimal path of a patrol inspection device; the prewarning module is used for generating a prewarning signal when a potential risk is detected; and the user interaction module is used for displaying the state of the system through an intuitive interface. Through the realization of real-time data processing, dynamic task scheduling and path optimization, the application improves the response speed and efficiency of the system, and enhances the equipment fault prevention and the decision-making ability of an operator.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence and Internet of Things, and particularly relates to a patrol service management system based on artificial intelligence and Internet of Things. BACKGROUND

[0002] In modern industry and urban management, it has become a common practice to use artificial intelligence and Internet of Things technology to dynamically monitor and maintain various facilities. These technologies enable remote monitoring of equipment status, optimization of operation and maintenance processes, and early prevention of potential equipment failures, especially in vast or difficult-to-access areas such as pipe networks, power lines, and large mechanical equipment. The efficient operation of the patrol service management system is crucial to ensuring the stability and safety of the facilities. However, although existing systems can collect a large amount of data and perform basic processing, they often lack effective task scheduling, path optimization, and real-time risk response capabilities.

[0003] The main problem of the prior art is that although basic data collection and monitoring can be performed, there is often a lack of effective data integration, intelligent task allocation, and early warning mechanisms, which limits the application efficiency of the patrol system in complex environments, especially in terms of response speed and accuracy when facing unexpected events and complex task priority adjustments. In addition, existing systems often do not provide sufficient user interaction functions, making it difficult for operators to make quick decisions and adjust patrol strategies based on real-time data.

[0004] Therefore, it is necessary to develop a new patrol service management system that can integrate advanced artificial intelligence algorithms and Internet of Things technology to achieve more advanced data processing, task scheduling, and early warning response, thereby improving the overall performance and adaptability of the system. SUMMARY

[0005] Based on the above purpose, the present application provides a patrol service management system based on artificial intelligence and Internet of Things.

[0006] The patrol service management system based on artificial intelligence and Internet of Things comprises a data acquisition module, an edge computing module, a task allocation module, a path planning module, an early warning module, and a user interaction module; wherein:

[0007] Data acquisition module: multiple sensors and mobile nodes deployed within the patrol area for real-time acquisition of multi-dimensional data of the environment and equipment, and transmission of the data to the edge computing module through Internet of Things protocol;

[0008] Edge computing module: receives multi-dimensional data from the data acquisition module and performs preliminary analysis and processing of the multi-dimensional data, including noise filtering and anomaly detection. The processed relevant information will be transmitted to the task allocation module and the path planning module;

[0009] Task allocation module: based on the relevant information processed by the edge computing module, generate inspection tasks according to the preset inspection rules, and dynamically optimize the priority and execution order of the tasks, and send the optimized task instructions to the path planning module for path generation;

[0010] Path planning module: according to the task instructions provided by the task allocation module and the environmental data processed by the edge computing module, use algorithms to generate the optimal path of the inspection equipment, and send the planning results to the inspection equipment for executing the inspection task;

[0011] Early warning module: receive the data processed by the edge computing module, combine historical data to predict and analyze the state of the equipment, when potential risks are detected, the early warning module generates an early warning signal, and at the same time feeds back the early warning information to the task allocation module to adjust the task plan;

[0012] User interaction module: used to display the running state, task progress and equipment state of the system through an intuitive interface, so that users can view the inspection data and adjust the inspection strategy, the user's adjusted instructions are conveyed to the task allocation module through the user interaction module, and the task allocation module updates the task execution strategy according to the adjustment.

[0013] Optionally, the data acquisition module includes a sensor unit, a mobile node unit and a data transmission unit; wherein:

[0014] Sensor unit: a plurality of sensors fixedly arranged in the inspection area, used for continuously monitoring environmental parameters and equipment states, the plurality of sensors including temperature sensors, humidity sensors, vibration sensors and image sensors, the temperature sensors being used for measuring temperature change data of the environment and the equipment, the humidity sensors being used for monitoring air humidity data, the vibration sensors being used for detecting mechanical vibration data of the equipment, and the image sensors being used for acquiring image information of the equipment and the environment;

[0015] Mobile node unit: a plurality of mobile nodes arranged in the inspection area, used for autonomously adjusting positions according to the requirements of the inspection tasks, so as to cover each monitoring point in the inspection area, the mobile node unit being equipped with an autonomous mobile device, used for dynamically monitoring the predetermined key area;

[0016] Data transmission unit: transmit the multi-dimensional data collected by the sensor unit and the mobile node unit to the edge computing module through an Internet of Things protocol.

[0017] Optionally, the edge computing module includes a data receiving unit and a data processing unit; wherein:

[0018] Data receiving unit: receiving multi-dimensional data from the data acquisition module through a high-speed wireless communication interface, and synchronously processing real-time data, and performing normalization processing after synchronous processing, and storing in a local cache in a standardized format;

[0019] Data processing unit: used for preliminary analysis and processing of data transmitted by the data receiving unit, and including a noise filtering subunit and an anomaly detection subunit;

[0020] Noise filtering subunit: adopting a multi-stage filtering algorithm, dynamically adjusting filtering strength according to preset filtering parameters while receiving data, to remove interference signals in environmental and equipment state data;

[0021] Anomaly detection subunit: used for real-time detection of anomalies in data, judging validity of data and filtering out abnormal data by analyzing mutations, overruns and unexpected changes.

[0022] Optionally, the noise filtering subunit includes first-stage filtering processing, second-stage filtering processing and third-stage filtering processing; the first-stage filtering processing outputs D1(t), the second-stage filtering processing outputs D2(t), and a dynamic strength adjustment mechanism is further arranged between the second-stage filtering processing and the third-stage filtering processing, and the dynamic strength adjustment mechanism is used for dynamically adjusting filtering strength according to changes in real-time acquisition data and preset filtering parameters P(t), and the adjusted filtering strength is represented as: Wherein, S(t) is an initial filtering strength, and α and β are filtering strength adjustment coefficients, S A (t) is the adjusted filtering strength, represents a change rate of the data D2(t) after second-stage filtering at time t; the third-stage filtering processing is performed after adjustment of the dynamic strength adjustment mechanism, and outputs D3(t), which is the final filtering output data.

[0023] Optionally, the anomaly detection subunit specifically includes:

[0024] Mutation analysis: performing mutation analysis on the final filtering output data D3(t), detecting rapid change patterns in data, and calculating a change rate V(t) of each data point from a previous time point;

[0025] Overrun analysis: checking whether data exceeds a normal range, and judging whether data is overrun by comparing the purified data D3(t) with preset upper and lower thresholds U L and L L , and specifically when D3(t)>U L or D3(t)<L L , the data point is marked as an overrun data point;

[0026] Expected deviation analysis: analyze the deviation ΔD(t) between the purified data and the expected model output to determine the validity of the data, calculate the deviation between the purified data D3(t) and the expected value D exp (t), the formula is: ΔD(t) = D3(t) - D exp (t), if the deviation ΔD(t) exceeds the set deviation threshold T Δ , it is marked as an unexpected data point;

[0027] Abnormal score calculation and determination: the results of mutation analysis, overrun analysis and expected deviation analysis are integrated to calculate the final abnormal score E f (t), the formula is: E f (t) = w1 × V(t) + w2 × overrun determination + w3 × ΔD(t), where w1, w2 and w3 are weight coefficients of each analysis method, if E f (t) exceeds the preset abnormal threshold T E , it is determined as abnormal data and filtered out;

[0028] Filtering abnormal data: for the data points determined as abnormal, remove them from the data set, and output the filtered data D f (t).

[0029] Optionally, the task allocation module includes a task generation unit, a priority evaluation unit, a sequence optimization unit and a task output unit; wherein:

[0030] Task generation unit: for receiving data D f (t) processed by the edge computing module, and generating specific inspection tasks according to the preset inspection rules R, the inspection rules R including device inspection frequency, environmental monitoring standards and safety requirements, by analyzing the rules R and combining the data D f (t), the task generation unit outputs the inspection task list T;

[0031] Priority evaluation unit: each task in the inspection task list T output by the task generation unit is evaluated for priority, the priority evaluation is based on multiple factors, including the importance I(T i ), urgency U(T i ) and device status S(T i ) of the task, the priority P(T i ) of each task T i ) is calculated by the formula: P(T i ) = w1 × I(T i ) + w2 × U(T i ) + w3 × S(T i ), where w1, w2, w3 are evaluation weight coefficients, P(T i) is the priority score of the ith task;

[0032] The sequence optimization unit sorts the task list T according to the task priority P(T i ) calculated by the priority evaluation unit, and the sequence optimization unit specifically uses a sorting function Sort(T, P(T i )) for sorting tasks according to the size of the task priority P(T i ), which is specifically represented as: Where O(T) is a task execution sequence list in descending order of priority, and tasks with high priority are executed first;

[0033] The task output unit transmits the optimized task execution sequence O(T) to the path planning module and the user interaction module.

[0034] Optionally, the path planning module includes a path calculation unit, an obstacle detection unit, and a path optimization unit; wherein:

[0035] The path calculation unit first parses the inspection target points and sequence in the task instruction, and combines the geographic information and device location in the environment data D f (t) to preliminarily calculate the path set P init of the inspection device;

[0036] The obstacle detection unit detects obstacles in the preliminary path set P init generated by the path calculation unit, identifies the obstacle positions on the inspection device path in combination with the environment data provided by the edge computing module, and excludes them from the preliminary path to form the adjusted path set P adj ;

[0037] The path optimization unit optimizes the adjusted path set P adj , adopts a path optimization strategy based on a heuristic algorithm to calculate the optimal path P opt of the inspection device, and the optimal path calculation considers multiple factors including the path length L(P j ), energy consumption E(P j ), and time T(P j ), and the formula of the optimal path is: P opt = min P (w1×L(P)+w2×E(P)+w3×T(P)), where w1, w2, and w3 are weight coefficients of path optimization, and min P represents that the path with the minimum objective function value among all paths P is the optimal path, and P opt is the optimal path;

[0038] The path output unit outputs the optimized optimal path P opt to the inspection device, so as to ensure that the device performs the inspection task according to the optimal path.

[0039] Optionally, the early warning module comprises a historical data comparison unit, a state prediction unit, an early warning signal generation unit and an early warning information feedback unit, wherein:

[0040] The historical data comparison unit receives the current data D f (t) processed by the edge computing module h (t), and compares the two to identify abnormal trends. The comparison is completed by calculating the deviation ΔD(t) of the current data and the historical data, and the formula is: ΔD(t) = D f (t) - D h (t). When ΔD(t) exceeds a set deviation threshold T D , it is recorded as an abnormal trend.

[0041] The state prediction unit predicts the future operating state of the device based on a regression algorithm. Let S p (t+Δt) be the predicted state result of the device. When S p (t+Δt) exceeds a preset safety threshold S safe , it is marked as a potential risk state.

[0042] The early warning signal generation unit is used to generate an early warning signal when the state prediction unit detects that the future state of the device has a potential risk. The early warning signal includes the risk level, the risk type and the recommended operation.

[0043] The early warning information feedback unit feeds back the information in the early warning signal to the task allocation module after generating the early warning signal. The task allocation module adjusts the task plan according to the risk level and type in the early warning signal, including prioritizing high-risk devices or rearranging the inspection path.

[0044] Optionally, the early warning information feedback unit specifically comprises:

[0045] The risk analysis receives the early warning signal transmitted by the early warning signal generation unit and analyzes the risk level R l and the risk type R t in the signal.

[0046] The task priority adjustment adjusts the priority of the task according to the risk level R l . The adjustment process is completed by recalculating the priority score P(T i ) of the task. The formula for the adjusted priority score is: P'(T i ) = P(T i ) + α × Rl wherein P'(T i ) is the adjusted task priority, P(T i ) is the original task priority, and a is the risk adjustment coefficient.

[0047] Task Reassignment: After the priority adjustment unit recalculates the priorities, the task reassignment unit reorders the tasks based on the new priority scores and reallocates resources.

[0048] Optionally, the user interaction module includes a state display unit, a task progress monitoring unit, a device state display unit, and a strategy adjustment unit.

[0049] State Display Unit: This unit is responsible for receiving real-time operational data from various modules and presenting them in a visual format, including dashboards, charts, and real-time status indicators.

[0050] Task Progress Monitoring Unit: This unit displays the progress of current inspection tasks in real-time, including the execution status of tasks, a list of completed tasks, ongoing tasks, and upcoming tasks.

[0051] Device State Display Unit: This unit displays the current state information of inspection devices, including their geographical location, operational status, battery level, sensor readings, and health status.

[0052] Strategy Adjustment Unit: This unit allows users to adjust inspection strategies based on real-time information about system operation, task progress, and device state.

[0053] Advantages of the Invention:

[0054] The present invention integrates advanced artificial intelligence and Internet of Things technology, significantly improving the dynamic monitoring capabilities of equipment and environmental conditions.

[0055] The application automatically adjusts the priority and order of inspection tasks through real-time data and risk assessment, and can also optimize the inspection path in real time according to environmental changes and equipment status. This dynamic adjustment mechanism ensures the optimal utilization of system resources and time management, greatly reducing operating costs. At the same time, the intuitive interface provided by the user interaction module allows the operating personnel to receive early warning information and task updates in real time, allowing them to quickly adjust the inspection strategy when necessary, enhancing the system's response capability to emergencies and ensuring the continuous and stable operation of key facilities. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only a part of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0057] Fig. 1 The schematic diagram of the inspection service management system of the embodiment of the application is shown in the figure.

[0058] Fig. 2 The schematic diagram of the edge computing module of the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0059] The application will be described in detail below in conjunction with the drawings and specific embodiments. It should be noted that in order to make the embodiments more detailed, the following embodiments are the best, preferred embodiments, and other alternative ways can also be used by those skilled in the art to implement them; and the drawings are only used to describe the embodiments in more detail, and are not intended to specifically limit the application.

[0060] It should be noted that in the specification, "one embodiment", "embodiment", "exemplary embodiment", "some embodiments" and the like indicate that the described embodiments can include specific features, structures or characteristics, but not necessarily every embodiment includes the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of those skilled in the related art to realize this feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0061] In general, terminology can be understood at least in part from the context of use. For example, the term "one or more" as used herein, depending at least in part upon context, can be used to describe any feature, structure, or characteristic in a singular sense or can be used to describe combinations of features, structures or characteristics in a plural sense. Additionally, the term, "based on," as used herein, can be understood as not necessarily of exclusive alternatives, but rather, alternatively, can permit that additional or

[0062] As shown in Figs. 1-2 The artificial intelligence and Internet of Things-based inspection service management system includes a data acquisition module, an edge computing module, a task allocation module, a path planning module, a warning module, and a user interaction module; wherein:

[0063] The data acquisition module: a variety of sensors (such as temperature sensors, humidity sensors, vibration sensors, image sensors) and mobile nodes deployed in the inspection area are used to collect multi-dimensional data of the environment and equipment in real time, and transmit the data to the edge computing module through the Internet of Things protocol for preliminary processing and analysis;

[0064] The edge computing module: receives the multi-dimensional data of the data acquisition module, and performs preliminary analysis and processing on the multi-dimensional data, including noise filtering and anomaly detection. The relevant information after processing will be transmitted to the task allocation module and the path planning module for subsequent task generation and path optimization. At the same time, when the network is unstable, the data processing task continues to be executed to ensure the continuity of the inspection;

[0065] The task allocation module: based on the relevant information processed by the edge computing module, the inspection tasks are generated according to the preset inspection rules, and the priority and execution order of the tasks are dynamically optimized. The optimized task instructions are sent to the path planning module for path generation, and are conveyed to the user through the user interaction module for reference and adjustment;

[0066] The path planning module: according to the task instructions provided by the task allocation module and the environmental data processed by the edge computing module, the optimal path of the inspection equipment is generated using an algorithm, and the planning result is sent to the inspection equipment for executing the inspection task;

[0067] The warning module: receives the data processed by the edge computing module, combines historical data to predict and analyze the state of the equipment, and generates a warning signal when a potential risk is detected. The warning information is fed back to the task allocation module to adjust the task plan;

[0068] User interaction module: used to display the running state, task progress and device status of the system through an intuitive interface, so that users can view the inspection data and adjust the inspection strategy, the user's adjusted instructions are transmitted to the task allocation module through the user interaction module, and the task allocation module updates the task execution strategy according to the adjustment.

[0069] The data acquisition module includes a sensor unit, a mobile node unit and a data transmission unit; wherein:

[0070] Sensor unit: a plurality of sensors fixedly arranged in the inspection area, used for continuously monitoring environmental parameters and device states, the plurality of sensors including temperature sensors, humidity sensors, vibration sensors and image sensors, the temperature sensors being used for measuring temperature change data of the environment and the device, the humidity sensors being used for monitoring air humidity data, the vibration sensors being used for detecting mechanical vibration data of the device, and the image sensors being used for acquiring image information of the device and the environment, each sensor collecting and processing monitoring data in real time through an internally integrated digital interface to ensure the accuracy and stability of the data;

[0071] Mobile node unit: a plurality of mobile nodes arranged in the inspection area, used for autonomously adjusting positions according to the needs of the inspection task to cover each monitoring point in the inspection area, the mobile node unit being equipped with an autonomous mobile device for dynamically monitoring the predetermined key area, and the mobile node unit further including a multi-sensor array for data acquisition to supplement the areas that cannot be covered by the fixed sensor unit;

[0072] Data transmission unit: transmitting the multi-dimensional data collected by the sensor unit and the mobile node unit to the edge computing module through an Internet of Things protocol, the data transmission unit integrating a high-speed wireless communication submodule, supporting real-time data transmission, and internally built with a data caching function to prevent data loss when the network is unstable, and the data transmission unit uses encryption technology to ensure the security and integrity of data transmission; through the above structure design, the data acquisition module can efficiently and accurately collect multi-dimensional data of the environment and the device in the inspection area, ensuring the comprehensiveness and real-time performance of the data, the fixed deployment of the sensor unit and the dynamic adjustment of the mobile node unit are combined to realize full coverage monitoring of the inspection area, and the data transmission unit ensures the safe transmission and timely processing of the data, improving the overall performance and reliability of the system.

[0073] The edge computing module includes a data receiving unit and a data processing unit; wherein:

[0074] Data receiving unit: receiving multi-dimensional data from the data acquisition module through a high-speed wireless communication interface, and synchronously processing real-time data to ensure that multi-dimensional data is not delayed and data is not lost during transmission. The received multi-dimensional data includes environmental parameters, device status information, and image data. After synchronous processing, normalization processing is performed, and the data is stored in a local cache in a standardized format for subsequent processing.

[0075] Data processing unit: used for preliminary analysis and processing of data transmitted by the data receiving unit. The data processing unit includes a noise filtering subunit and an anomaly detection subunit.

[0076] Noise filtering subunit: uses a multi-stage filtering algorithm to dynamically adjust the filtering strength based on preset filtering parameters while receiving data, to remove interference signals in environmental and device status data, ensuring data purity and accuracy.

[0077] Anomaly detection subunit: used for real-time detection of anomalies in data. By analyzing mutations, overruns, and unexpected changes, the validity of the data is determined and abnormal data is filtered out to improve the accuracy of system judgment. Through the above design, the edge computing module can efficiently and accurately process the received multi-dimensional data in real time. The combination of the noise filtering subunit and the anomaly detection subunit ensures the reliability and purity of the data, providing a solid data foundation for subsequent task allocation and path planning, and improving the stability and response speed of the system as a whole.

[0078] The noise filtering subunit includes first-level filtering processing, second-level filtering processing, and third-level filtering processing. The first-level filtering processing outputs D1(t), and the second-level filtering processing outputs D2(t). A dynamic strength adjustment mechanism is also provided between the second-level filtering processing and the third-level filtering processing. The dynamic strength adjustment mechanism is used to dynamically adjust the filtering strength based on the changes in the real-time collected data and the preset filtering parameters P(t). The adjusted filtering strength is represented as: where S(t) is the initial filtering strength, and a and b are filtering strength adjustment coefficients. S A (t) is the adjusted filtering strength, represents the change rate of the data D2(t) after second-level filtering at time t. The dynamic strength adjustment mechanism adjusts the third-level filtering processing to output D3(t), which is the final filtered data.

[0079] The processing steps of noise filtering are as follows:

[0080] Data initialization: receiving raw data transmitted from the data receiving unit and performing preliminary preprocessing on the data. The preprocessed data is represented as D(t), where D(t) represents the multi-dimensional data set received at time t.

[0081] Pre-set filter parameter loading: according to the running environment and the device state, load the applicable filter parameter P(t) from the pre-set filter parameter set, represented as: P(t)={P1(t), P2(t),..., P n (t)}, wherein P i (t) represents the i-th filter parameter used at time t;

[0082] Primary filter processing: use a low-pass filter algorithm to preliminarily filter the original data, filter out high-frequency noise, and the processed data is represented as D1(t), the formula is: D1(t)=D(t)-H(D(t), P L (t)), wherein (D(t), P L (t)) represents the filtering operation on the original data D(t) according to the low-pass filter parameter P L (t), P L (t) is the low-pass filter parameter used at time t;

[0083] Secondary filter processing: perform band-pass filtering on the data D1(t) after low-pass filtering to remove noise in a predetermined frequency band, and the processed data is represented as D2(t), the formula is: D2(t)=D1(t)-B(D1(t), P B (t)), wherein B(D1(t), P B (t)) represents the band-pass filtering operation according to the band-pass filter parameter P B (t), P B (t) is the band-pass filter parameter used at time t;

[0084] Dynamic intensity adjustment: dynamically adjust the intensity of the filter according to the changes in the real-time collected data and the pre-set filter parameter P(t);

[0085] Tertiary filter processing: apply the dynamically adjusted filter intensity S A (t) to the data for tertiary filter processing to remove subtle noise, and finally obtain the purified data D3(t), represented as: D3(t)=D2(t)-S A (t)×G(D2(t)), wherein G(D2(t)) represents the filtering operation according to the dynamic intensity adjustment, and D3(t) is the final filter output data;

[0086] Output purified data: take the final filtered data D3(t) as the output data for processing by the anomaly detection unit; through the above steps, the noise filtering sub-unit can dynamically adjust the filter intensity according to the real-time situation while receiving data, remove the interference signals in the environmental and device state data in stages and multiple levels, ensure the high purity and accuracy of the output data, and provide a reliable data basis for subsequent anomaly detection and system judgment.

[0087] The anomaly detection subunit specifically includes:

[0088] Mutation analysis: Perform mutation analysis on the final filtered output data D3(t), detect rapid change patterns in the data, calculate the change rate V(t) of each data point from the previous time point, and compare it with the preset mutation threshold T V , expressed as: Where Δt is the time interval, when V(t)>T V , it is marked as a mutation data point;

[0089] Exceeding limit analysis: Check if the data exceeds the normal range, compare the purified data D3(t) with the preset upper limit U L and lower limit L L threshold, judge whether the data exceeds the limit, specifically when D3(t)>U L or D3(t)<L L , it is marked as an exceeding limit data point;

[0090] Expected deviation analysis: Analyze the deviation ΔD(t) between the purified data and the expected model output to determine the validity of the data, calculate the deviation between the purified data D3(t) and the expected value D exp (t), the formula is: ΔD(t)=D3(t)-D exp (t), if the deviation ΔD(t) exceeds the set deviation threshold T Δ , it is marked as a data point that does not meet the expectations;

[0091] Anomaly score calculation and determination: The results of mutation analysis, exceeding limit analysis and expected deviation analysis are integrated to calculate the final anomaly score E f (t), the formula is: E f (t)=w1×V(t)+w2×exceeding limit determination+w3×ΔD(t), where w1, w2 and w3 are weight coefficients of each analysis method, if E f (t) exceeds the preset anomaly threshold T E , it is determined as an abnormal data and is filtered out;

[0092] Filtering abnormal data: For data points determined as abnormal, remove them from the data set and output the filtered data D f (t) for subsequent processing by the system; Through the above steps, the anomaly detection subunit can effectively identify mutations, exceeding limits and unexpected changes in the data, calculate the anomaly score comprehensively, and accurately filter out abnormal data. This multi-dimensional anomaly detection method ensures the effectiveness and reliability of the system input data, and provides a high-quality data basis for further decision-making and control.

[0093] The task allocation module comprises a task generation unit, a priority evaluation unit, a sequence optimization unit and a task output unit; wherein:

[0094] The task generation unit is configured to receive the data D f (t) processed by the edge computing module, and generate specific inspection tasks according to a preset inspection rule R, the inspection rule R comprising a device inspection frequency, an environmental monitoring standard and a safety requirement, the rule R being parsed and combined with the data D f (t) to generate the inspection task list T, which is represented as: T = {T1, T2,..., Ti,...}. n} wherein Ti is the i-th inspection task generated according to the rule R. i

[0095] The rule parsing process comprises the following steps:

[0096] Rule loading and initialization: first, the preset inspection rule R is loaded from a system configuration file or a database, the rule R comprising a series of parameters and conditions for describing the device inspection frequency, the environmental monitoring standard, the safety requirement and the like;

[0097] Data matching and condition judgment: for each rule R i , the task generation unit extracts relevant data from the purification data D f (t) received from the edge computing module for matching, specifically, the task generation unit compares the conditions in each rule with the real-time data to determine whether the current data satisfies the specific inspection condition, the formula being: if D f (t)[k]≥R i [k] then the rule is satisfied, wherein D f (t)[k] represents the k-th data dimension in the purification data related to the rule R i .

[0098] Rule priority analysis: the rule R comprises multiple conditions, and there is a priority relationship between different conditions, the task generation unit analyzes the weight of each condition according to the rule priority, and preferentially satisfies the rules with high priority, for example, if the priority of the rule R1 is higher than that of the rule R2, the condition satisfying the R1 will be preferentially considered when generating the task;

[0099] Task generation: according to the rule analysis result, the task generation unit creates the inspection task, if the data D f (t) satisfies a rule R i , the corresponding inspection task T i is generated.

[0100] ​Priority evaluation unit: each task in the inspection task list T output by the task generation unit is evaluated in priority, the priority evaluation is based on multiple factors, including the importance I(T i ), the urgency U(T i ) and the device state S(T i ) of the task, and the priority P(T i ) of each task T i ) is calculated according to the formula: P(T i ) = w1 x I(T i ) + w2 x U(T i ) + w3 x S(T i ), wherein w1, w2, w3 are evaluation weight coefficients, and P(T i ) is the priority score of the i-th task;

[0101] Sequence optimization unit: according to the task priority P(T i ) calculated by the priority evaluation unit, the task list T is sorted, and the sequence optimization unit specifically uses the sorting function Sort(T, P(T i )) to sort the tasks according to the size of the task priority P(T i ), which is specifically represented as: Wherein O(T) is a task execution sequence list arranged in descending order of priority, and tasks with high priority are executed first;

[0102] Task output unit: the optimized task execution sequence O(T) is transmitted to the path planning module and the user interaction module to ensure that the system executes the inspection tasks in the optimal order, and at the same time, the task information is displayed to the user for monitoring and adjustment; through the above design, the task allocation module can automatically generate inspection tasks meeting the requirements based on the data processed by the edge computing module and the preset inspection rules, and dynamically optimize the priority and execution order of the tasks according to the importance, urgency and device state of the tasks. The multi-level optimization mechanism of this module ensures efficient allocation and execution of inspection tasks, and improves the response speed and overall performance of the system.

[0103] The path planning module includes a path calculation unit, an obstacle detection unit and a path optimization unit; wherein:

[0104] Path calculation unit: first analyze the inspection target points and sequence in the task instruction, and combine the geographic information and device location in the environment data D f (t) to preliminarily calculate the path set P init of the inspection device, which is represented as: P init = {P1, P2,..., P m}, wherein P iThis represents the i-th path that the inspection equipment needs to access. The initial path is generated based on the geographical distribution of the equipment location and the task target point; the specific formula is:

[0105] P init ={P i |P i =Path(S, T) i D f (t)), i = 1, 2, ..., m}, where P i This represents the i-th path that the inspection equipment needs to access, where S is the starting position of the equipment, and T is the starting position of the equipment. i D represents the target location in the mission instructions. f (t) represents the environmental data at the current time t, and the function Path(S, T) i D f (t) represents the distance from the starting position S to the target position T. i The path calculation results;

[0106] Obstacle detection unit: The initial path set P generated by the path calculation unit. init Obstacle detection is performed to ensure the safety and accessibility of the inspection path. The obstacle detection unit, combined with environmental data provided by the edge computing module, identifies the location of obstacles on the inspection equipment's path and removes them from the initial path, forming an adjusted path set P. adj ;

[0107] Path optimization unit: for the adjusted path set P adj To optimize the path, a path optimization strategy based on heuristic algorithms (such as A* algorithm and Dijkstra's algorithm) is adopted to calculate the optimal path P for the inspection equipment. opt The optimal path calculation considers multiple factors, including the path length L(P). j Energy consumption E (P) j ) and time T(P j The formula for the optimal path is: P opt =min P (w1×L(P)+w2×E(P)+w3×T(P)), where w1, w2, and w3 are the weight coefficients for path optimization, min P The optimal path is the path P that minimizes the objective function among all paths. opt The optimal path is one that meets the task requirements while minimizing path costs.

[0108] Path output unit: Outputs the optimized path P optThe transmission to the inspection device ensures that the device performs the inspection task according to the optimal path. Through the above design, the path planning module can generate the optimal path most suitable for the inspection device according to the task instructions provided by the task allocation module and the environmental data processed by the edge computing module. Through multi-level path calculation, obstacle detection and path optimization, the module ensures that the inspection device can efficiently and safely complete the inspection task in a complex environment, while optimizing the path cost and improving the overall operation efficiency of the system.

[0109] The early warning module includes a historical data comparison unit, a state prediction unit, a warning signal generation unit and a warning information feedback unit; wherein:

[0110] The historical data comparison unit receives the current data D f (t) processed by the edge computing module h (t) and compares them to identify abnormal trends. The comparison is completed by calculating the deviation ΔD(t) of the current data and the historical data, and the formula is: ΔD(t) = D f (t) - D h (t). When ΔD(t) exceeds the set deviation threshold T D , it is recorded as an abnormal trend for subsequent state prediction.

[0111] The state prediction unit predicts the future operating state of the device based on a regression algorithm. Let S p (t+Δt) be the predicted state result of the device. When S p (t+Δt) exceeds the preset safety threshold S safe , it is marked as a potential risk state.

[0112] The specific steps of prediction are as follows:

[0113] Step 1: Obtain the current data D f (t), the historical deviation data ΔD(t) and the historical operating data D h (t) from the historical data comparison unit. Standardize these data to ensure that the data input is in the same dimension. The standardization formula is: where μ D and σ D are the mean and standard deviation of the data set D(t), respectively, and D norm (t) is the standardized data.

[0114] Step 2: Input the standardized data D norm (t) into the feature extraction submodule to extract the key feature vector X(t) as the input of the regression model. The features include historical trends, current state and deviation information.

[0115] Step 3: Establishing a device state prediction model using a linear regression algorithm, the training process of the model is based on the historical data set D h (t) and the corresponding actual device state S(t), the trained model is used to predict the device state S p (t+Δt) at future time t+Δt, the regression prediction formula is: Where β0 is the intercept term of the model, β i is the regression coefficient of the model, X i (t) is the i-th feature in the input feature vector;

[0116] Step 4: The calculated device prediction state S p (t+Δt) is used as output for subsequent warning signal generation, if S p (t+Δt) exceeds the safety threshold S safe , it is marked as a potential risk state.

[0117] Warning signal generation unit: used to generate a warning signal when the state prediction unit detects that the future state of the device has potential risks, the warning signal includes risk level, risk type and recommended operation, and is sent to the user interaction module and task allocation module through the communication interface, so that appropriate measures can be taken in time;

[0118] Warning information feedback unit: after generating the warning signal, the warning information feedback unit feeds back the information in the warning signal to the task allocation module, which will adjust the task plan according to the risk level and type in the warning signal, including prioritizing high-risk devices or rearranging the inspection path, to ensure the safe operation of the system; through the above design, the warning module can effectively combine current data and historical data to predict the device state, and generate a warning signal in time when potential risks are detected, and feed back to the task allocation module to adjust the task plan, thereby improving the safety and response speed of the system and ensuring the reliable execution of the inspection task.

[0119] The warning information feedback unit specifically includes:

[0120] Risk analysis: receiving the warning signal transmitted by the warning signal generation unit, analyzing the risk level R l and risk type R t in the signal, the risk level R l is a quantitative indicator in the warning signal, used to represent the severity of the risk, the range is set as R l ∈[1, 5], where 1 represents low risk and 5 represents high risk, the risk type R t is a classification of specific risk categories, including device failure and environmental anomalies;

[0121] Task priority adjustment: according to risk level R l The priority of the task is adjusted, and the adjustment process is completed by recalculating the priority score P(T i ) of the task. The formula for the adjusted priority score is: P'(T i ) = P(T i ) + α × R l , where P'(T i ) is the adjusted task priority, P(T i ) is the original task priority, and α is the risk adjustment coefficient, which quantifies the impact of risk level on task priority. The higher the risk level, the greater the priority adjustment, ensuring that high-risk tasks are given priority.

[0122] Task reallocation: After the task priority adjustment unit recalculates the priority, the task reallocation unit reorders the tasks according to the new priority score and reallocates resources. Through the above design, the early warning information feedback unit can effectively analyze the risk level and type in the early warning signal and convert the risk information into the basis for adjusting the task priority. After receiving the adjusted priority information, the task allocation module dynamically adjusts the inspection plan through accurate task reordering and resource allocation, ensuring that the system can respond promptly when dealing with high-risk events, optimizing resource allocation, and improving the response speed and task execution efficiency of the inspection system.

[0123] The user interaction module includes a state display unit, a task progress monitoring unit, a device state display unit, and a strategy adjustment unit. Among them:

[0124] State display unit: used to receive real-time running data from various modules and present it in a visual way, including dashboards, charts, and real-time updated status indicators. Users can view key parameters such as current system running load, network connection status, and data processing through the interface, thereby comprehensively mastering the system's running situation.

[0125] Task progress monitoring unit: real-time display of the progress of the current inspection task, including the execution status of the task, the list of completed tasks, the tasks in progress, and the tasks about to start. The task progress monitoring unit also includes the execution order of the tasks through the form of a timeline or Gantt chart, and distinguishes different task states (such as not started, in progress, completed, and pending adjustment) through color coding, helping users intuitively understand the progress of task execution.

[0126] The device state display unit is used to show the current state information of the inspection device, including the geographical position, running state, battery capacity, sensor reading and health state of the device. The device state display unit also includes the position information and state of each device, which is presented to the user in the form of a map and state icon, so that the user can click on the specific device icon to view detailed information, such as sensor data, device historical operation record, etc.

[0127] The policy adjustment unit allows the user to adjust the inspection policy according to the real-time information of the system running state, task progress and device state. The policy adjustment unit provides an intuitive operation interface, and the user can adjust the priority of the task, modify the execution order of the task or add a new inspection task by dragging and dropping, clicking or inputting instructions. The policy adjustment unit transmits the user's adjustment instructions to the task allocation module, which updates the inspection plan in real time according to the user's instructions, and feeds back the updated information to the user interaction module for display.

[0128] The present application encompasses any alternative, modification, equivalent method and scheme made on the essence and scope of the present application. In order to make the public have a thorough understanding of the present application, specific details are described in the following preferred embodiments of the present application, and the present application can also be fully understood without the description of these details to those skilled in the art. In addition, in order to avoid unnecessary confusion to the essence of the present application, well-known methods, processes, procedures, elements and circuits, etc. are not described in detail.

[0129] The above is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered as the protection scope of the present application.

Claims

1. An inspection service management system based on artificial intelligence and the Internet of Things, characterized in that, The system comprises a data acquisition module, an edge computing module, a task allocation module, a path planning module, a warning module and a user interaction module. The data acquisition module: a plurality of sensors and mobile nodes deployed in the inspection area are used to collect multi-dimensional data of the environment and equipment in real time, and transmit the data to the edge computing module through the Internet of Things protocol. The edge computing module: receives multi-dimensional data from the data acquisition module, and performs preliminary analysis and processing on the multi-dimensional data, including noise filtering and anomaly detection. The processed information will be transmitted to the task allocation module and the path planning module. The edge computing module comprises a data receiving unit and a data processing unit. The data receiving unit: receives multi-dimensional data from the data acquisition module through a high-speed wireless communication interface, and synchronously processes the real-time data. After synchronous processing, the data is normalized and stored in the local cache in a standardized format. The data processing unit: used for preliminary analysis and processing of the data transmitted by the data receiving unit. The data processing unit comprises a noise filtering subunit and an anomaly detection subunit. The noise filtering subunit: uses a multi-stage filtering algorithm to dynamically adjust the filtering strength according to the preset filtering parameters while receiving the data, to remove interference signals in the environmental and equipment state data. The anomaly detection subunit: used for real-time detection of anomalies in the data. By analyzing mutations, overruns and unexpected changes, the effectiveness of the data is judged and abnormal data is filtered out. The noise filtering subunit comprises a first filtering process, a second filtering process and a third filtering process; the first filtering process outputs , the second filtering process outputs , and a dynamic intensity adjustment mechanism is further arranged between the second filtering process and the third filtering process, which is used to dynamically adjust the filtering intensity according to the change of the real-time collected data and the preset filtering parameters , the adjusted filtering intensity is represented as: , wherein, is an initial filtering intensity, and are filtering intensity adjustment coefficients, is the adjusted filtering intensity, represents the data after the second filtering, is the change rate of the data at time ; the third filtering process is performed after the adjustment of the dynamic intensity adjustment mechanism, and outputs , which is the final filtering output data. The task allocation module: based on the relevant information processed by the edge computing module, generates inspection tasks according to the preset inspection rules, and dynamically optimizes the priority and execution order of the tasks. The optimized task instructions are sent to the path planning module for path generation. The path planning module: generates the optimal path of the inspection equipment using algorithms based on the task instructions provided by the task allocation module and the environmental data processed by the edge computing module, and sends the planning results to the inspection equipment for execution of the inspection task. The warning module: receives the data processed by the edge computing module, combines historical data to predict and analyze the state of the equipment, and generates a warning signal when potential risks are detected. The warning information is fed back to the task allocation module to adjust the task plan. The user interaction module: used to display the running state, task progress and equipment state of the system through an intuitive interface, so that users can view the inspection data and adjust the inspection strategy. The user-adjusted instructions are transmitted to the task allocation module through the user interaction module, and the task allocation module updates the task execution strategy according to the adjustment. 2.The AI and IoT-based inspection service management system according to claim 1, wherein, The data acquisition module comprises a sensor unit, a mobile node unit and a data transmission unit. The sensor unit: a plurality of sensors fixedly deployed in the inspection area are used to continuously monitor environmental parameters and equipment states. The plurality of sensors comprise temperature sensors, humidity sensors, vibration sensors and image sensors. The temperature sensors are used to measure temperature change data of the environment and equipment, the humidity sensors are used to monitor air humidity data, the vibration sensors are used to detect mechanical vibration data of the equipment, and the image sensors are used to acquire image information of the equipment and environment. Mobile node unit: multiple mobile nodes are arranged in the inspection area, which can autonomously adjust the position according to the demand of the inspection task, so as to cover each monitoring point in the inspection area, and the mobile node unit is equipped with an autonomous moving device for dynamic monitoring of the predetermined key area; Data transmission unit: transmit the multi-dimensional data collected by the sensor unit and the mobile node unit to the edge computing module through the Internet of Things protocol. 3.The AI and IoT-based inspection service management system of claim 1, wherein, The anomaly detection subunit specifically comprises: Mutation analysis: on the final filtered output data Perform mutation analysis, detect fast changing patterns in the data, calculate the rate of change of each data point from the previous time point ; Out of limit analysis: check if the data is out of normal range, by comparing the cleaned data with pre-set upper limit and lower limit threshold, determine if the data is out of limit, specifically when or , then mark as out of limit data point; Expected deviation analysis: by analyzing the deviation of the sanitized data from the expected model output Determining the validity of the data, calculating sanitized data The deviation between the expected value is calculated as follows: If the deviation exceeds the set deviation threshold , it is marked as an unexpected data point; Abnormal score calculation and determination: the results of mutation analysis, ultra-limit analysis and expected deviation analysis are integrated to calculate the final abnormal score , the formula is: wherein, , and are the weight coefficients of each analysis method, if exceeds the preset abnormal threshold , it is determined as abnormal data and filtered. Filtering of abnormal data: For data points determined to be abnormal, remove them from the dataset and output the filtered data . 4.The AI and IoT-based inspection service management system of claim 3, wherein, The task allocation module comprises a task generation unit, a priority evaluation unit, a sequence optimization unit and a task output unit; wherein: Task generation unit: for receiving data processed by the edge computing module and generating specific inspection tasks according to preset inspection rules , the inspection rules including equipment inspection frequency, environmental monitoring standards and safety requirements, by analyzing rules and combining data , the task generation unit outputs an inspection task list ; Priority assessment unit: performs priority assessment on each task in the list of patrol tasks output by the task generation unit, the priority assessment being based on a plurality of factors including the importance , urgency and device status of each task , the priority of each task being calculated according to the formula: wherein is the assessment weight coefficient, is the priority score of the th task. The sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit , and the sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit , and the sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit , and the sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit , and the sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit , and the sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit , and the sequence optimization unit sorts the tasks according to the priority of the tasks calculated by the priority evaluation unit Task output unit: the optimized task execution order is transmitted to the path planning module and the user interaction module. 5.The AI and IoT-based inspection service management system of claim 1, wherein The path planning module comprises a path calculation unit, an obstacle detection unit and a path optimization unit; wherein: Path computation unit: first resolve the inspection target point and sequence in the task instruction, and combine the geographic information and device position in the environment data to preliminarily calculate the path set of the inspection device ; Obstacle detection unit: preliminary path set generated by path calculation unit Obstacle detection is performed in combination with the environmental data provided by the edge computing module to identify the position of obstacles on the inspection device path and exclude them from the preliminary path to form an adjusted path set ; path optimization unit: the adjusted path set is optimized, a path optimization strategy based on a heuristic algorithm is adopted to calculate the optimal path of the inspection device The optimal path calculation considers multiple factors, including path length , energy consumption and time , the formula of the optimal path is: , wherein and are the weight coefficients of path optimization, indicates the path that minimizes the objective function value among all paths is the optimal path, is the optimal path; path output unit: output the optimized optimal path to the inspection device, ensuring that the device performs the inspection task according to the optimal path. transmit to the inspection device, ensuring that the device performs the inspection task according to the optimal path. 6.The AI and IoT-based inspection service management system according to claim 1, wherein, The early warning module comprises a historical data comparison unit, a state prediction unit, a warning signal generation unit and a warning information feedback unit; wherein: Historical data comparison unit: receives current data processed from edge computing module and historical running data of the device , compares the two to identify abnormal trends, compares the deviation of current data from historical data , and the formula is: When the set deviation threshold is exceeded , it is recorded as an abnormal trend; State prediction unit: based on regression algorithm to predict the future running state of the equipment, set Predicted state results for the equipment, when ) exceeds the preset safety threshold , then marked as a potential risk state; The early warning signal generation unit is used for generating an early warning signal when the state prediction unit detects that the future state of the equipment has potential risks, and the early warning signal comprises a risk level, a risk type and a recommended operation; The early warning information feedback unit feeds back the information in the early warning signal to the task allocation module after generating the early warning signal, and the task allocation module adjusts the task plan according to the risk level and type in the early warning signal, including preferentially processing high-risk equipment or rearranging the inspection path. 7.The AI and IoT-based inspection service management system according to claim 6, wherein, The early warning information feedback unit specifically comprises: Risk analysis: receiving the early warning signal transmitted by the early warning signal generation unit, analyzing the risk level and risk type in the signal and risk type ; Task priority adjustment: according to risk level Adjust the priority of the task, the adjustment process by recalculating the priority score of the task Complete, the adjusted priority score formula is: Where, The adjusted task priority, The original task priority, The risk adjustment coefficient; Task reallocation: after the task priority adjustment unit recalculates the priority, the task reallocation unit reorders the tasks according to the new priority score and reallocates resources. 8.The AI and IoT-based inspection service management system of claim 1, wherein, The user interaction module comprises a state display unit, a task progress monitoring unit, a device state display unit and a strategy adjustment unit; wherein: The state display unit is used for receiving running data from each module in real time and presenting in a visual manner, including a dashboard, a chart and a real-time updated state indicator light; The task progress monitoring unit displays the progress of the current inspection task in real time, including the execution state of the task, the completed task list, the ongoing task and the upcoming task, and the task progress monitoring unit also comprises a time axis or a Gantt chart to display the execution sequence of the task and distinguish different task states by color coding; The device state display unit is used for displaying the current state information of the inspection device, including the geographical position, the running state, the battery capacity, the sensor reading and the health state of the device; The strategy adjustment unit allows users to adjust the inspection strategy according to the real-time information of the system running state, the task progress and the device state, and the strategy adjustment unit provides an intuitive operation interface, and users can adjust the priority of the task, modify the execution sequence of the task or add a new inspection task by dragging, clicking or inputting instructions.

Images