A financial voucher generation method, medium and system based on a marine monitoring platform

Abstract

The application provides a financial voucher generation method, medium and system based on a marine monitoring platform, belongs to the technical field of financial voucher generation methods, and collects operation state data, environment parameter data and equipment measurement data of marine monitoring equipment to establish an equipment operation database, performs time sequence change characteristic analysis to construct a floating change matrix, obtains a comprehensive error relationship matrix through Bayesian model analysis, decomposes to obtain an error stability matrix and an error change matrix, inputs the error change matrix into a marine equipment error classifier for classification, establishes a marine asset value evaluation model to predict equipment value, automatically identifies financial business events through a financial business event identification rule library, automatically matches accounting subjects according to the event type and error weight coefficient to generate a financial voucher template, and automatically enters the financial system after compliance verification of the financial voucher template, thereby solving the technical problem that the financial accounting result and the real value change of the equipment do not match.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of financial voucher generation methods, and in particular, relates to a financial voucher generation method based on a marine monitoring platform, a medium and a system. BACKGROUND

[0002] In the field of financial management of marine monitoring equipment, the traditional financial voucher generation method mainly relies on a pre-set fixed template and manual operation of accountants, and handles the financial accounting of equipment assets through a standardized depreciation calculation method and regular maintenance cost allocation. This method has been widely used in the management of conventional equipment in land environments. However, the traditional technology has obvious defects when dealing with marine monitoring equipment. Due to the complexity and harshness of the marine environment, the actual running state, error characteristics and value changes of the equipment show a high degree of dynamics and uncertainty, and the traditional fixed template cannot capture these real-time change characteristics, resulting in a serious disconnection between the generated financial vouchers and the real state of the equipment. In the current financial management of marine monitoring equipment, due to the lack of accurate evaluation mechanism for the actual running error and environmental loss of the equipment, the traditional method cannot accurately reflect the real value change rule of the equipment in the marine environment, that is, the existing technology has the technical problem that the generation of financial vouchers is often based on manual experience and fixed templates, and cannot be well adjusted dynamically according to the actual running state and error characteristics of the marine monitoring equipment, resulting in a mismatch between the financial accounting results and the real value change of the equipment. SUMMARY

[0003] Therefore, the application provides a financial voucher generation method based on a marine monitoring platform, a medium and a system, which can solve the technical problem that the generation of financial vouchers in the prior art is often based on manual experience and fixed templates, and cannot be well adjusted dynamically according to the actual running state and error characteristics of the marine monitoring equipment, resulting in a mismatch between the financial accounting results and the real value change of the equipment.

[0004] The application is implemented in the following manner: a financial voucher generation method based on a marine monitoring platform is provided in the first aspect of the application, which comprises the following steps: collecting the operation state data, environmental parameter data and equipment measurement data of the marine monitoring equipment in real time through Internet of Things sensors to establish an equipment operation database, recording and maintaining historical data, and dynamically adjusting the data collection frequency; extracting the equipment measurement data from the equipment operation database to analyze the time sequence change characteristics and construct a floating change matrix, analyzing the floating change matrix through a Bayesian model to obtain a comprehensive error relationship matrix, performing singular value decomposition on the comprehensive error relationship matrix to obtain an error stability matrix and an error change matrix, inputting the error change matrix into a marine equipment error classifier to obtain an error classification result, and calculating the error weight coefficient through a gated weight function according to the error classification result; performing state analysis and abnormality detection on the operation state data in combination with the error classification result, calculating the equipment health index and the expected maintenance cost; calculating the dynamic depreciation rate and the spare parts demand; establishing a marine asset value evaluation model to calculate the equipment value evaluation result; automatically identifying the financial business event according to the equipment health index change, the expected maintenance cost, the equipment value evaluation result and the error weight coefficient; automatically matching the corresponding accounting subject according to the financial business event by calling the accounting subject mapping rule, generating the financial voucher template and entering the financial system.

[0005] In the step of calculating the dynamic depreciation rate and the spare parts demand, the environmental wear coefficient is calculated based on the equipment health index and the expected maintenance cost in combination with the operation intensity and the corrosion index, the dynamic depreciation rate is calculated by using the workload method in combination with the environmental wear coefficient, and the spare parts demand is determined according to the fault frequency, the expected maintenance cost and the error weight coefficient.

[0006] The data collection frequency adjustment coefficient is a frequency adjustment parameter calculated according to the operation intensity and the corrosion index, which is used to dynamically adjust the data collection interval of the Internet of Things sensors.

[0007] The floating change matrix is a dynamic change characteristic matrix constructed by mathematical modeling of the time sequence change characteristics of the equipment measurement data, which reflects the fluctuation law of the equipment measurement accuracy.

[0008] The Bayesian model analysis process comprises the following steps: taking the measurement data change mode in the floating change matrix as the observation evidence, establishing the prior probability distribution of systematic error, random error, environmental interference error and drift error, calculating the posterior influence probability of each type of error on different monitoring parameters through Bayesian inference method, and constructing the comprehensive error relationship matrix containing the association strength between the error type and the monitoring parameter.

[0009] The marine equipment error classifier is an error type identification model constructed based on deep learning technology, which can automatically identify and classify different types of equipment measurement errors, and output error classification results including probability distribution of each error type and final classification label.

[0010] The calculation method of the gating weight function is based on the confidence score of each error type in the error classification result. When the system error confidence exceeds 0.8, the corresponding error weight coefficient is set to 1.5; when the random error confidence exceeds 0.7, the corresponding error weight coefficient is set to 0.8; and when the environmental interference error confidence exceeds 0.6, the corresponding error weight coefficient is dynamically adjusted between 1.0 and 1.3 according to the corrosive index.

[0011] The marine asset value evaluation model is an asset value prediction model constructed based on time series data analysis method, taking running intensity, corrosive index, failure frequency, expected maintenance cost, error stability matrix and error change matrix as input parameters, processing error stability matrix and error change matrix through error feature fusion layer, and predicting equipment remaining service life through time series data analysis.

[0012] The running state data includes equipment usage time, running intensity, and failure frequency. The environmental parameter data includes seawater temperature, salinity, pressure, and corrosive index. The equipment measurement data includes real-time measurement values of each monitoring parameter.

[0013] In the step of calculating the equipment health index, the equipment health index is calculated according to the equipment usage time, failure frequency, corrosive index and error weight coefficient, and the expected maintenance cost is calculated according to the equipment health index and maintenance history data.

[0014] The singular value decomposition is a mathematical method for decomposing the comprehensive error relationship matrix into error stability matrix and error change matrix, which is used to separate systematic error and random error components.

[0015] The equipment health index is a device state evaluation index calculated by comprehensively considering equipment usage time, failure frequency, corrosive index and error weight coefficient, which is used to evaluate the current running state of the equipment.

[0016] The environmental wear coefficient is an adjustment parameter for environmental correction of standard depreciation rate according to corrosive index and running intensity. The dynamic depreciation rate is a depreciation rate of the equipment calculated by using workload method combined with environmental wear coefficient.

[0017] The construction process of the marine asset value evaluation model comprises collecting historical equipment operation data from an equipment operation database as training samples, segmenting the historical equipment operation data according to a time window, each training sample comprising operation state data, environmental parameter data and historical error weight coefficients for 30 consecutive days as input features, and the corresponding actual depreciation amount and residual value of the equipment as label data.

[0018] The financial business event comprises a depreciation event, a maintenance cost event, a spare part procurement event and an asset impairment event, and the error weight coefficient is used to adjust the identification priority of the financial business event.

[0019] The accounting subject mapping rule is a configuration rule for automatically associating the financial business event with the corresponding accounting subject, and the financial voucher template is a standardized financial voucher format generated according to the financial business event type, the equipment value evaluation result, the dynamic depreciation rate, the spare part demand and the error weight coefficient.

[0020] The second aspect of the present application provides a computer readable storage medium, the computer readable storage medium stores program instructions, the program instructions are run in the computer, and the program instructions are used for executing the financial voucher generation method based on the marine monitoring platform.

[0021] The third aspect of the present application provides a financial voucher generation system based on a marine monitoring platform, comprising the computer readable storage medium, the system is any one of a computer, a server and a single chip microcomputer, the computer readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing the program instructions stored in the computer readable storage medium.

[0022] The present application solves the technical problem that the traditional financial accounting method cannot accurately reflect the real value change of the marine monitoring equipment by establishing a real-time data acquisition system based on an Internet of Things sensor and a marine equipment error classifier, combining a Bayesian model analysis and a singular value decomposition technology to construct a dynamic financial voucher generation mechanism. The present application adopts an error feature fusion and a dynamic depreciation rate calculation method, can automatically adjust a financial processing strategy according to the actual operation state of the equipment, environmental parameters and error weight coefficients, overcomes the technical defect that the traditional fixed template method has poor adaptability in the marine environment, and realizes accurate matching between the financial voucher generation and the real state of the equipment. The present application establishes a scientific financial business event identification rule library by comprehensively considering the equipment health index, the environmental loss coefficient and the error classification result, and technically ensures that the generated financial voucher can accurately reflect the real value change trend of the marine monitoring equipment in the complex marine environment, and solves the core technical problem that the financial accounting result does not match the actual state of the equipment. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 A flowchart of the method of the present application. DETAILED DESCRIPTION

[0024] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0025] As Figure 1 shown is a flowchart of a financial voucher generation method based on a marine monitoring platform according to a first aspect of the present application, and the method comprises the following steps:

[0026] S01, real-time collection of operation state data, environmental parameter data and equipment measurement data of marine monitoring equipment through Internet of Things sensors, establishment of an equipment operation database, wherein the operation state data comprises equipment usage time length, operation intensity and fault frequency, the environmental parameter data comprises seawater temperature, salinity, pressure and corrosion index, and the equipment measurement data comprises real-time measurement values of various monitoring parameters, and maintenance history data is recorded in the equipment operation database, a data collection frequency adjustment coefficient is calculated according to the operation intensity and the corrosion index, and the data collection frequency is dynamically adjusted through the data collection frequency adjustment coefficient;

[0027] S02, extraction of the equipment measurement data from the equipment operation database, time sequence change characteristic analysis of the equipment measurement data to construct a floating change matrix, error analysis of the marine monitoring equipment, analysis of the floating change matrix through a Bayesian model to obtain a comprehensive error relationship matrix, singular value decomposition of the comprehensive error relationship matrix to obtain an error stability matrix and an error change matrix, input of the error change matrix into a marine equipment error classifier for classification to obtain an error classification result, and calculation of an error weight coefficient through a gated weight function according to the error classification result;

[0028] S03, extraction of the operation state data, the environmental parameter data and the maintenance history data from the equipment operation database, state analysis and abnormality detection of the operation state data in combination with the error classification result, identification of equipment operation abnormality, maintenance demand and fault early warning, calculation of an equipment health degree index according to the equipment usage time length, the fault frequency, the corrosion index and the error weight coefficient, and calculation of an expected maintenance cost according to the equipment health degree index and the maintenance history data;

[0029] S04, based on the equipment health index and the expected maintenance cost, combined with the operation intensity and the corrosion index, calculate the environmental wear coefficient, use the workload method to calculate the dynamic depreciation rate combined with the environmental wear coefficient, determine the spare parts demand according to the failure frequency, the expected maintenance cost and the error weight coefficient;

[0030] S05, establish a marine asset value evaluation model, take the operation intensity, the corrosion index, the failure frequency, the expected maintenance cost, the error stability matrix and the error change matrix as input parameters, process the error stability matrix and the error change matrix through the error feature fusion layer, predict the equipment remaining service life combined with time series data analysis, and calculate the equipment value evaluation result combined with the dynamic depreciation rate;

[0031] S06, establish a financial business event identification rule library, automatically identify financial business events according to the equipment health index change, the expected maintenance cost, the equipment value evaluation result and the error weight coefficient, adjust the identification priority of the financial business events through the error weight coefficient, and the financial business events include equipment depreciation events, maintenance cost events, spare parts procurement events and asset impairment events;

[0032] S07, according to the event type of the financial business event, the equipment value evaluation result and the error weight coefficient, call the accounting subject mapping rule to automatically match the corresponding accounting subject, and generate a financial voucher template combined with the dynamic depreciation rate and the spare parts demand;

[0033] S08, perform compliance verification on the financial voucher template, verify the integrity and accuracy of the voucher elements through the rule engine, automatically enter the financial system and generate audit trail records according to the financial business events.

[0034] Among them, the data collection frequency adjustment coefficient is a frequency adjustment parameter calculated according to the operation intensity and the corrosion index, which is used to dynamically adjust the data collection interval of the Internet of Things sensor.

[0035] Among them, the maintenance history data is the historical record information of the equipment historical maintenance operation, maintenance cost and maintenance effect recorded in the equipment operation database.

[0036] Among them, the floating change matrix is a dynamic change feature matrix constructed by mathematical modeling of the time series change characteristics of the equipment measurement data, reflecting the fluctuation law of the measurement accuracy of the equipment.

[0037] Among them, the error analysis is the evaluation and analysis of the measurement accuracy of the marine monitoring equipment, and the error includes systematic error, random error, environmental interference error and drift error.

[0038] The comprehensive error relationship matrix is a multi-dimensional correlation matrix obtained by analyzing the relationship between various errors and monitoring parameters through a Bayesian model, describing the influence degree and interaction relationship of different error types on each monitoring parameter.

[0039] The singular value decomposition is a mathematical method for decomposing the comprehensive error relationship matrix into the error stability matrix and the error variation matrix, used to separate systematic error and random error components.

[0040] The error stability matrix is a matrix reflecting the characteristics of systematic errors obtained by singular value decomposition, representing the long-term stable error mode of the device.

[0041] The error variation matrix is a matrix reflecting the characteristics of random errors obtained by singular value decomposition, representing the dynamic error mode of the device.

[0042] The marine equipment error classifier is an error type identification model constructed based on deep learning technology, which can automatically identify and classify different types of device measurement errors.

[0043] The error classification result is the error type identification and corresponding confidence score output by the marine equipment error classifier after classification processing of the error variation matrix.

[0044] The gating weight function is a function for calculating weight coefficients based on the confidence scores of each error type in the error classification result, used to quantify the influence degree of different error types on device operation.

[0045] The error weight coefficient is a weight parameter calculated by the gating weight function according to the error classification result, used to adjust the importance of error influence in subsequent analysis.

[0046] The device health index is a device state evaluation index calculated by combining the device usage time, the failure frequency, the corrosion index and the error weight coefficient, used to evaluate the current operating state of the device.

[0047] The environmental wear coefficient is an adjustment parameter for environmental correction of the standard depreciation rate according to the corrosion index and the operating intensity.

[0048] The dynamic depreciation rate is a depreciation rate of the device calculated by using the workload method combined with the environmental wear coefficient, reflecting the actual value loss speed of the device in the marine environment.

[0049] The spare parts demand quantity is a spare parts reserve quantity calculated according to the failure frequency, the expected maintenance cost and the error weight coefficient, and is used for formulating a spare parts procurement plan.

[0050] The error feature fusion layer is a processing layer of the marine asset value evaluation model for processing the error stability matrix and the error change matrix, and is used for fusing error features with equipment operation features.

[0051] The marine asset value evaluation model is an asset value prediction model constructed based on a time series data analysis method, and predicts equipment value by analyzing the running intensity, the corrosion index, the failure frequency, the expected maintenance cost, the error stability matrix and the error change matrix.

[0052] The equipment value evaluation result is equipment current value and future value change prediction data output by the marine asset value evaluation model.

[0053] The financial business event is a business activity that needs to be processed according to the equipment health index change, the expected maintenance cost, the equipment value evaluation result and the error weight coefficient.

[0054] The accounting subject mapping rule is a configuration rule for automatically associating the financial business event with a corresponding accounting subject, and ensures the accuracy of financial voucher generation.

[0055] The financial voucher template is a standardized financial voucher format generated according to the financial business event type, the equipment value evaluation result, the dynamic depreciation rate, the spare parts demand quantity and the error weight coefficient.

[0056] The rule engine is a software component for automatic verification and processing based on predefined business rules, and is used for verifying the compliance of the financial voucher template.

[0057] The audit track record is complete trace information recorded from the whole process of collecting the running state data to generating the financial voucher template.

[0058] The construction process of the marine asset value evaluation model includes collecting historical equipment operation data from the equipment operation database as training samples, the historical equipment operation data including the equipment usage time, the operation intensity, the failure frequency, the corrosion index, the maintenance history data and the corresponding equipment actual value change record, segmenting the historical equipment operation data according to a time window, each training sample including the operation state data, the environment parameter data and historical error weight coefficients for 30 consecutive days as input features, the corresponding equipment actual depreciation amount and residual value as label data, fusing the error stability matrix and the error change matrix with the equipment operation features in the error feature fusion layer through a weighted fusion method, and training the marine asset value evaluation model by using a regression analysis method, so that the model can predict the equipment value evaluation result according to the current operation intensity, the corrosion index, the failure frequency, the expected maintenance cost, the error stability matrix and the error change matrix.

[0059] The Bayesian model analysis process includes taking the measurement data change pattern in the floating change matrix as observation evidence, establishing the prior probability distribution of systematic error, random error, environmental interference error and drift error, calculating the posterior influence probability of each type of error on different monitoring parameters by Bayesian inference method, and constructing the comprehensive error relationship matrix including the association strength of error types and monitoring parameters. The row of the comprehensive error relationship matrix represents different error types, the column represents different monitoring parameters, and the matrix element value represents the influence strength of the corresponding error type on the corresponding monitoring parameter.

[0060] The training process of the marine equipment error classifier includes collecting historical error data of marine monitoring equipment from the equipment operation database as training samples, the historical error data including the error change matrix of labeled error types, constructing a deep neural network model for error pattern learning, training the marine equipment error classifier to identify systematic error, random error, environmental interference error and drift error by a supervised learning method, and outputting the error classification result including the probability distribution of each error type and the final classification label.

[0061] The calculation method of the gating weight function is based on the confidence score of each error type in the error classification result. When the system error confidence exceeds 0.8, the corresponding error weight coefficient is set to 1.5 to increase the influence weight of systematic problems. When the random error confidence exceeds 0.7, the corresponding error weight coefficient is set to 0.8 to reduce the influence weight of random fluctuations. When the environmental interference error confidence exceeds 0.6, the corresponding error weight coefficient is dynamically adjusted between 1.0 and 1.3 according to the corrosion index. When the drift error confidence exceeds 0.5, the corresponding error weight coefficient is set to 1.2 and the device calibration suggestion is started.

[0062] The financial business event identification rule library contains 4 types of event identification rules. The device depreciation event identification rule automatically triggers the generation of depreciation vouchers according to the dynamic depreciation rate and the depreciation period. The maintenance cost event identification rule triggers the generation of expense vouchers when the expected maintenance cost exceeds the budget threshold and the error weight coefficient is greater than 1.0. The spare parts procurement event identification rule triggers the generation of procurement vouchers when the spare parts demand exceeds the inventory safety line and the error weight coefficient indicates a device failure risk. The asset impairment event identification rule triggers the generation of impairment vouchers when the device health index is below 0.3 and the error weight coefficient indicates an irreparable error. The error weight coefficient is used to adjust the trigger conditions of each type of financial business event, ensuring that business events with high error impact are prioritized.

[0063] The specific implementation of the above steps is described in detail below.

[0064] The specific implementation of step S01 is to collect comprehensive data from marine monitoring devices through an Internet of Things sensor network and dynamically adjust them. First, a distributed sensor network is established. Each sensor node collects running state data such as device usage time, running intensity, and failure frequency at preset time intervals. The device usage time records the cumulative working time in hours, the running intensity uses a normalized value of 0 to 1 to represent the device load degree, and the failure frequency records the number of failures per unit time. Environmental parameter data such as seawater temperature, salinity, pressure, and corrosion index are also collected. The seawater temperature monitoring range is -2℃ to 35℃, the salinity monitoring range is 0 to 45‰, the pressure monitoring range is 0 to 6× Pa, corrosive index is represented by a quantitative scale of 0 to 10. The sensor also collects instantaneous measurement values of each monitoring parameter in real time as equipment measurement data. The time, cost and effect of each maintenance operation are recorded synchronously in the equipment operation database as maintenance history data. To realize dynamic optimization of data acquisition frequency, the system calculates a data acquisition frequency adjustment coefficient according to the product of operation intensity and corrosive index. When the product is less than 2, the adjustment coefficient is set to 0.5; when the product is between 2 and 6, the adjustment coefficient is set to 1.0; and when the product is greater than 6, the adjustment coefficient is set to 1.5. The basic acquisition frequency is adjusted by a corresponding multiple through this adjustment coefficient to ensure that the data acquisition density is increased in high-load or high-corrosion environments.

[0065] The specific implementation of step S02 is to perform deep error analysis and intelligent classification on equipment measurement data. The system extracts equipment measurement data from the equipment operation database, uses time series analysis method to extract time sequence change characteristics of the data, and constructs measurement value changes in continuous time period into a floating change matrix through sliding window technology, which reflects the dynamic fluctuation pattern of equipment measurement accuracy. Then, multi-dimensional error analysis is performed on the marine monitoring equipment to identify four main error types, namely system error, random error, environmental interference error and drift error. The floating change matrix is analyzed by using Bayesian inference method to establish the probability association relationship between each type of error and monitoring parameter, and a comprehensive error relationship matrix is generated. The matrix is decomposed by using singular value decomposition algorithm to separate the complex error relationship into error stability matrix reflecting systematic error characteristics and error change matrix reflecting random error characteristics. The error change matrix is sent as input data to the marine equipment error classifier based on deep neural network, which identifies different error types through multi-layer perception structure and outputs confidence score. According to the confidence of each error type in the error classification result, the error weight coefficient is calculated by using a gating weight function. When the system error confidence exceeds 0.8, the weight coefficient is set to 1.5; when the random error confidence exceeds 0.7, the weight coefficient is set to 0.8; when the environmental interference error confidence exceeds 0.6, the weight coefficient is dynamically adjusted between 1.0 and 1.3; and when the drift error confidence exceeds 0.5, the weight coefficient is set to 1.2.

[0066] The specific implementation of step S03 is to comprehensively operate state data, environmental parameter data and maintenance history data to evaluate the health status of the equipment. The system extracts relevant data from the equipment operation database, combines error classification results to perform in-depth state analysis on the operation state data, uses statistical analysis methods to identify abnormal patterns in the data, and detects equipment operation abnormalities by setting threshold values. When the equipment usage time exceeds 80% of the designed service life, a service life warning is triggered; when the failure frequency exceeds 150% of the average value, a failure warning is triggered; when the corrosion index exceeds 7, an environmental warning is triggered. The system uses a multi-parameter fusion algorithm to calculate the equipment health index, which takes into account the time decay factor of equipment usage time, the reliability influence factor of failure frequency, the environmental damage factor of corrosion index and the precision influence factor of error weight coefficient. Through weighted average method, the health score in the range of 0 to 1 is obtained. Based on the equipment health index and maintenance history data, a regression analysis method is used to predict future maintenance needs, and the expected maintenance cost is calculated through historical maintenance cost trend analysis and equipment health degree change trend, which includes preventive maintenance cost and failure repair cost.

[0067] The specific implementation of step S04 is to calculate dynamic economic parameters based on the health status of the equipment and environmental factors. The system calculates the environmental wear coefficient based on the equipment health index and the expected maintenance cost, combined with the operation intensity and the corrosion index, which reflects the accelerated impact of the marine environment on the wear of the equipment value. The environmental wear coefficient is calculated using a nonlinear combination of operation intensity and corrosion index. When the product of the two is less than 3, the wear coefficient is 1.0; when the product is between 3 and 8, the wear coefficient increases linearly from 1.2 to 1.8; when the product exceeds 8, the wear coefficient is 2.0. The dynamic depreciation rate is calculated using the workload method, which determines the depreciation progress based on the ratio of actual workload to designed workload, and modifies the standard depreciation rate based on the environmental wear coefficient, so that the depreciation rate can truly reflect the value decay speed of the equipment in the marine environment. The spare parts demand is determined based on the failure frequency, the expected maintenance cost and the error weight coefficient, and the safety stock model in inventory management theory is used. When the failure frequency exceeds 0.1 times per day, the spare parts demand increases by 20%; when the expected maintenance cost exceeds 5% of the original value of the equipment, the spare parts demand increases by 30%; when the error weight coefficient is greater than 1.2, the spare parts demand increases by an additional 15%.

[0068] The specific implementation of step S05 is to build a comprehensive marine asset value assessment model for equipment value prediction. The model takes operating intensity, corrosion index, failure frequency, expected maintenance cost, error stability matrix and error change matrix as core input parameters. The error feature fusion layer in the model uses attention mechanism to process error stability matrix and error change matrix, and highlights the influence of important error features on asset value through adaptive weight distribution. The fusion layer fuses error matrix features and equipment operation features at multiple levels to form a comprehensive feature vector. The model uses time series prediction algorithm to analyze the historical trend of equipment operation data and predict the remaining service life of the equipment. This prediction is based on the degradation model of equipment health index and the cumulative damage model of environmental factors. Combined with the dynamic depreciation rate and equipment remaining life prediction, the model outputs the current value assessment and future value trend of the equipment. The value assessment results include comprehensive assessment of three dimensions of book value, fair value and recoverable value.

[0069] The specific implementation of step S06 is to establish an intelligent financial business event identification system. The system establishes a financial business event identification rule library containing four categories of event identification rules, and each category of rule is optimized and adjusted in combination with error weight coefficients. The equipment depreciation event identification rule is automatically triggered based on the dynamic depreciation rate and the pre-designed design review period, and the priority is increased when the monthly depreciation rate exceeds 2%. The maintenance cost event identification rule monitors the change of expected maintenance cost, and automatically triggers the generation of cost voucher demand when the cost exceeds the budget threshold by 10% and the error weight coefficient is greater than 1.0. The spare parts procurement event identification rule compares the demand quantity of spare parts with the current inventory, and triggers the procurement voucher demand when the demand quantity exceeds the inventory safety line by 30% and the error weight coefficient shows that the equipment failure risk increases. The asset impairment event identification rule is based on the equipment health index monitoring, and triggers the impairment voucher demand when the health degree is less than 0.3 and the error weight coefficient shows that there is an unrepairable error. The system dynamically adjusts the identification priority of various financial business events through the error weight coefficient, ensures that business events with high error influence degree are given priority, and improves the timeliness and accuracy of financial processing.

[0070] The specific implementation of step S07 is to realize automatic matching of accounting subjects and generation of financial voucher templates based on a business rule engine. The system calls pre-configured accounting subject mapping rules for automatic matching based on the identified financial business event type, which is established based on enterprise accounting standards and industry practices. The equipment depreciation event corresponds to the fixed asset depreciation subject, the maintenance cost event corresponds to the management cost or operating cost subject, the spare parts procurement event corresponds to the raw material procurement subject, and the asset impairment event corresponds to the asset impairment loss subject. The system determines the voucher amount in combination with the equipment value assessment result, and adjusts the amount in precision using an error weight coefficient. When the error weight coefficient is greater than 1.2, an uncertainty adjustment of 5% is added to the related amount. The depreciation amount is calculated in combination with the dynamic depreciation rate, and the procurement amount is determined according to the demand for spare parts, to generate a standardized financial voucher template containing accounting subjects, debit and credit directions, amounts, and summaries. The voucher template also contains auxiliary information such as the time of business occurrence, related equipment number, and error impact rating, to ensure the completeness and traceability of the financial voucher.

[0071] The specific implementation of step S08 is to ensure the quality of financial vouchers and complete system integration through multi-level compliance verification. The system uses a rule engine based on pre-defined business rules to comprehensively verify the financial voucher template, including verification of the completeness of voucher elements, verification of debit and credit balance, verification of the reasonableness of the amount, and verification of the consistency of business logic. The completeness verification ensures that the voucher contains necessary accounting subjects, amounts, and summary information, the debit and credit balance verification ensures that the debit amount is equal to the credit amount, the amount reasonableness verification ensures that the amount is within a reasonable range by comparing with historical data, and the business logic verification ensures that the voucher content is correct in relation to the logical relationship of the triggering business event. After passing the verification, the system automatically enters the financial voucher into the enterprise financial management system, and uses a standard interface protocol to ensure the security and accuracy of data transmission. At the same time, the system generates a complete audit trail record based on the financial business event, which contains information from data collection, error analysis, event identification to voucher generation. The record records the timestamp, operation content, related parameters and processing results of each processing link, to ensure the completeness and traceability of the financial processing process.

[0072] The detailed structure of the Bayesian model includes four core components: prior probability distribution establishment layer, evidence processing layer, posterior probability calculation layer, and relationship matrix generation layer. The prior probability distribution establishment layer establishes the initial probability distribution for systematic error, random error, environmental interference error, and drift error based on historical error statistical data. The prior probability of systematic error adopts normal distribution, the random error adopts Gaussian distribution, the environmental interference error adopts Weibull distribution, and the drift error adopts exponential distribution. The evidence processing layer takes the measurement data change pattern in the floating change matrix as the observation evidence, and processes the input data through three sub-processes of data preprocessing, feature extraction, and evidence quantification. The posterior probability calculation layer uses Bayesian theorem for inference calculation, calculates the influence probability of each type of error on different monitoring parameters through the likelihood function, and obtains the posterior probability distribution combined with the prior probability. The relationship matrix generation layer converts the posterior probability into matrix form, and constructs a comprehensive error relationship matrix with rows representing error types and columns representing monitoring parameters. The training data set of the model includes collecting historical running data of at least 1000 marine monitoring devices for more than 6 months, which contains the measurement values and corresponding true values of monitoring parameters such as temperature, salinity, pressure, and dissolved oxygen. Through manual annotation, the types of errors are identified, and the annotation data set of the error type and monitoring parameter influence relationship is established. The cross-validation method is used to train the model parameters to ensure the generalization ability of the model in different sea areas and different device types.

[0073] The detailed structure of the marine equipment error classifier adopts a deep convolutional neural network architecture, including four main modules: input layer, feature extraction layer, classification decision layer, and output layer. The input layer receives the error change matrix obtained by singular value decomposition as input data, with a matrix size of n x m, where n represents the time dimension and m represents the monitoring parameter dimension. The feature extraction layer contains three convolutional layers and two pooling layers. The first convolutional layer uses 32 3 x 3 convolutional kernels to extract local features, the second convolutional layer uses 64 3 x 3 convolutional kernels to extract deep features, and the third convolutional layer uses 128 3 x 3 convolutional kernels to extract high-level features. Each convolutional layer is followed by a batch normalization layer and a ReLU activation function, and the pooling layer uses 2 x 2 max pooling to reduce the feature dimension. The classification decision layer contains two fully connected layers. The first fully connected layer contains 256 neurons, and the second fully connected layer contains 4 neurons corresponding to four error types, with a Softmax activation function outputting a probability distribution. The output layer generates error classification results, including confidence scores for each error type and a final classification label. The training dataset for this classifier requires the collection of at least 5000 error change matrix samples, each annotated by a marine engineer for error type. The data sources cover error patterns under different sea areas, different equipment models, and different environmental conditions, with an 8:1:1 ratio for training set, validation set, and test set. The model is trained using the stochastic gradient descent optimization algorithm, with a learning rate of 0.001, a batch size of 32, and a training period of 100 rounds. The model performance is evaluated by accuracy, precision, recall, and F1 score.

[0074] The detailed structure of the marine asset value evaluation model adopts a multi-input deep regression network architecture, including five core modules: a multi-path input layer, an error feature fusion layer, a time series analysis layer, a value prediction layer, and an output layer. The multi-path input layer receives scalar inputs such as operating intensity, corrosion index, failure frequency, and expected maintenance cost, and matrix inputs such as error stability matrix and error change matrix. The error feature fusion layer uses an attention mechanism to process error matrix inputs, calculates the importance weights of matrix elements through a self-attention module, and fuses the feature information of error stability matrix and error change matrix through a cross-attention module. The fused features are connected with the scalar inputs to form a comprehensive feature vector. The time series analysis layer uses a long short-term memory network to process time series features, including an LSTM layer with 128 hidden units to capture the time dependence of device operating state, and a bidirectional LSTM to enhance the model's ability to model historical and future trends. The value prediction layer contains three fully connected layers, with 256 neurons in the first layer, 128 neurons in the second layer, and 64 neurons in the third layer. Each layer uses a ReLU activation function and Dropout regularization technique to prevent overfitting. The output layer contains three output nodes corresponding to the predicted results of device book value, fair value, and recoverable value. The training dataset for this model requires at least 800 marine monitoring devices with complete life cycle data for more than 2 years. The data includes device purchase price, maintenance records, operating state data, environmental parameter data, and final disposal value, etc. The corresponding relationship between input features and asset value changes is established, and the time window sliding method is used to generate training samples. Each sample contains 30 consecutive days of input features and corresponding value change labels. The mean squared error is used as the loss function, the Adam optimization algorithm is used to train the model, the learning rate is set to 0.001, the batch size is set to 16, the training period is set to 200 rounds, and the model prediction accuracy is evaluated by mean absolute error, root mean square error, and coefficient of determination.

[0075] It should be noted that first, the present application adopts a Bayesian model to analyze the time sequence change characteristics of the measurement data of the marine monitoring equipment, and captures the fluctuation law of the measurement accuracy of the equipment by constructing a floating change matrix, which has a significant technical advantage compared with the traditional static error analysis method. The traditional method usually uses a fixed error correction coefficient or a simple linear regression model to process equipment errors, which cannot effectively identify and distinguish the complex error types and their interaction relationships in the marine environment. The present application establishes the prior probability distribution of systematic errors, random errors, environmental interference errors and drift errors by Bayesian inference method, can accurately calculate the posterior influence probability of various errors on different monitoring parameters, and constructs a multi-dimensional associated comprehensive error relationship matrix. Combined with singular value decomposition technology, the comprehensive error relationship matrix is decomposed into error stability matrix and error change matrix, realizing the accurate separation of systematic error and random error components, providing a reliable mathematical foundation for subsequent error classification and weight calculation. This error analysis method based on probability reasoning can effectively handle the uncertainty and complexity of equipment errors in the marine environment.

[0076] Second, the present application constructs a marine equipment error classifier based on deep learning technology, which can automatically identify and classify different types of equipment measurement errors, and has higher accuracy and consistency compared with the traditional artificial experience judgment method. The traditional method relies on the subjective experience of technicians to judge the error type and influence degree of the equipment, which has strong subjectivity, poor consistency and cannot handle complex error patterns. The present application collects historical error data of marine monitoring equipment as training samples, uses supervised learning method to train error classifier to identify systematic errors, random errors, environmental interference errors and drift errors, and outputs the probability distribution and confidence score of each error type. Based on the error classification result, the gate weight function can dynamically calculate the error weight coefficient according to the confidence score of different error types. When the confidence of systematic error is high, the influence weight of systematic problem is increased, and when the confidence of random error is high, the influence weight of random fluctuation is reduced. This dynamic weight adjustment mechanism can ensure the accurate matching of financial processing strategy and actual error state of the equipment.

[0077] Thirdly, the application establishes a financial business event recognition rule library, and dynamically adjusts the recognition priority of the financial business event through an error weight coefficient, which has stronger adaptability and accuracy compared with the traditional fixed template financial processing method. The traditional method uses a preset fixed template and a standardized depreciation calculation method, and cannot dynamically adjust according to the actual running state and error characteristics of the equipment, so that the generated financial voucher is out of touch with the real value change of the equipment. The application automatically recognizes the equipment depreciation event, maintenance cost event, spare part procurement event and asset impairment event according to the equipment health index change, expected maintenance cost, equipment value evaluation result and error weight coefficient, adjusts the triggering conditions of various financial business events through the error weight coefficient, and ensures that the business events with high error influence degree are processed preferentially. The financial voucher template is generated in combination with the dynamic depreciation rate and the demand for spare parts, so that the real-time association between the financial voucher generation and the actual state of the equipment is realized, and the dynamic adjustment mechanism based on the error characteristics can accurately reflect the real value change trend of the marine monitoring equipment in the complex environment.

[0078] The synergistic effect of the above three key technical ideas constitutes a complete dynamic financial voucher generation system, which has significant overall advantages compared with the traditional static financial processing method. The Bayesian model error analysis and singular value decomposition technology provide high-quality input data for the error classifier, ensuring the accuracy and reliability of error recognition; the dynamic adjustment mechanism of the marine equipment error classifier and the gating weight function provides scientific weight parameters for the financial voucher generation, realizing the accurate matching between the financial processing strategy and the actual state of the equipment; and the dynamic financial voucher generation mechanism based on the error weight coefficient converts the achievements of the first two technical ideas into a specific financial processing scheme, forming a closed-loop automated processing flow. The three technical ideas support and promote each other, and together build an intelligent financial management system driven by real-time data, which can automatically adapt to the complex changes of the running state of the equipment in the marine environment, solve the limitations of the traditional fixed template method in processing the financial accounting of the marine monitoring equipment, and realize the intelligentization, dynamicization and precision of the financial voucher generation.

[0079] The second aspect of the application provides a computer readable storage medium, the computer readable storage medium stores program instructions, the program instructions are used to execute the above-mentioned financial voucher generation method based on the marine monitoring platform when running in the computer.

[0080] The third aspect of the application provides a financial voucher generation system based on a marine monitoring platform, which comprises the above-mentioned computer readable storage medium, the system is any one of a computer, a server and a single-chip microcomputer, the computer readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing the program instructions stored in the computer readable storage medium.

[0081] Specifically, the principle of this invention is as follows: The fundamental principle that enables this invention to solve the problems of existing technologies lies in the construction of a dynamic financial voucher generation system based on real-time data-driven approaches. This system collects real-time operational status data, environmental parameter data, and equipment measurement data from marine monitoring equipment through IoT sensors, establishing a complete equipment operation database and providing a reliable data foundation for subsequent error analysis and value assessment. The technical solution employs a Bayesian model to analyze the temporal variation characteristics of equipment measurement data. By constructing a floating variation matrix and a comprehensive error relationship matrix, it can accurately identify and quantify systematic errors, random errors, environmental interference errors, and drift errors of equipment in the marine environment. Furthermore, through singular value decomposition (SVD), the errors are separated into stable and variable matrices, providing a mathematical basis for error classification and weight calculation. The core innovation of this invention lies in the establishment of a marine equipment error classifier and a gating weight function. This function can dynamically calculate error weight coefficients based on the error classification results and integrate these weight coefficients into the equipment health index, dynamic depreciation rate, and financial business event identification process, achieving real-time correlation between financial processing strategies and the actual status of the equipment. By using a marine asset valuation model and a financial business event identification rule base, this invention can automatically identify equipment depreciation events, maintenance cost events, spare parts procurement events, and asset impairment events. It also adjusts the priority of event identification based on error weight coefficients to ensure that the generated financial vouchers can accurately reflect the real value changes of equipment in the marine environment. This dynamic adjustment mechanism based on error characteristics and real-time status is the key to solving the limitations of traditional fixed template methods.

[0082] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0083] The specific implementation of step S01 involves comprehensive data collection and dynamic adjustment of marine monitoring equipment through an Internet of Things (IoT) sensor network. The calculation of the data collection frequency adjustment coefficient is specifically expressed as follows:

[0084] ;

[0085] In the formula, This is a coefficient for adjusting the data acquisition frequency. The value represents the equipment operating intensity, ranging from 0 to 1. The corrosion index ranges from 0 to 10. It is a piecewise function.

[0086] The specific piecewise function is defined as follows:

[0087] ;

[0088] In the formula, for The product value.

[0089] The parameter acquisition method is as follows: The method involves real-time monitoring using sensors, including step 1: monitoring the operating current of the equipment using a current sensor; step 2: monitoring the operating load of the equipment using a load sensor; and step 3: normalizing the ratio of the monitored values ​​to the rated parameters of the equipment to obtain the operating intensity. Data is obtained through environmental monitoring, including step 1: via Sensors monitor seawater pH levels. Step 1: Hydrogen ion concentration index; Step 2: Monitor seawater conductivity using a conductivity sensor; Step 3: Monitor seawater oxygen content using a dissolved oxygen sensor; Step 4: Quantify multiple parameters into a comprehensive level from 0 to 10 according to the corrosion evaluation standard.

[0090] The specific implementation of step S02 involves performing in-depth error analysis and intelligent classification of the equipment measurement data. The construction of the floating change matrix is ​​specifically represented as follows:

[0091] ;

[0092] In the formula, It is a floating change matrix; For the first Time of the first The change in each monitoring parameter; The time window length is measured in hours and is determined based on the equipment monitoring frequency; it is generally between 24 and 168 hours. The number of monitoring parameters.

[0093] The change is expressed as follows:

[0094] ;

[0095] In the formula, For the first Time of the first The measured values ​​of each monitoring parameter.

[0096] The calculation of posterior probability in Bayesian model analysis is specifically represented as follows:

[0097] ;

[0098] In the formula, For the first Type 1 error on the first The posterior influence probability of each monitoring parameter; In the first Under the error condition, the first The likelihood probability of each monitoring parameter is determined by the statistical relationship between error type and monitoring parameter deviation in historical data; The prior probability of the first error type is determined according to historical error statistics data, and the prior probabilities of the system error, the random error, the environmental interference error, and the drift error are respectively 0.3, 0.4, 0.2, and 0.1; The prior probability of the first error type is determined according to historical error statistics data, and the prior probabilities of the system error, the random error, the environmental interference error, and the drift error are respectively 0.3, 0.4, 0.2, and 0.1; The marginal probability of the first monitoring parameter is calculated by summing the probabilities of all error types. The marginal probability of the first monitoring parameter is calculated by summing the probabilities of all error types.

[0099] The calculation of the comprehensive error relationship matrix is specifically represented as follows:

[0100] ;

[0101] In the formula, is the comprehensive error relationship matrix; respectively represent the system error, the random error, the environmental interference error, and the drift error; represents the first monitoring parameter. represents the first monitoring parameter.

[0102] The singular value decomposition process is specifically represented as follows:

[0103] ;

[0104] In the formula, is the left singular matrix; is the singular value diagonal matrix; is the transpose of the right singular matrix.

[0105] The extraction of the error stability matrix and the error change matrix is represented as follows:

[0106] ;

[0107] ;

[0108] In the formula, is the error stability matrix; is the error change matrix; is the dimension number of the stable error component, which is determined by the cumulative contribution rate, and the value of is determined when the cumulative contribution rate of the current singular value reaches 85%; represents the first column of the matrix; represents the first row of the diagonal matrix; represents the first row and column submatrix of the diagonal matrix; represents the first row of the matrix; represents the first row of the matrix; represents the first the first column of the matrix to the last column of the matrix; denotes the first row of the matrix to the last row and the last column of the matrix; denotes the first row to the last row of the matrix.

[0109] The calculation of the gating weight function is specifically represented as follows:

[0110] ;

[0111] wherein, is an error weight coefficient; is a weight factor of the i-th class error; is a confidence score of the i-th class error; is a gating activation function.

[0112] The gating activation function is defined as:

[0113] ;

[0114] wherein, the parameter acquisition method is: obtained by using the marine equipment error classifier output, including the following steps: step 1: inputting the error change matrix into the classifier; step 2: calculating by forward propagation of the deep learning model; and step 3: generating the confidence of each class error by the output layer Softmax function. The default value of is determined according to the importance of the influence of the error type on the equipment operation, wherein , the system error has the greatest influence and the highest weight, and the random error has the least influence and the lowest weight.

[0115] The specific implementation of step S03 is to evaluate the health state of the equipment by comprehensively considering the operation state data, the environmental parameter data and the maintenance history data. The calculation of the equipment health index is specifically represented as follows:

[0116] ;

[0117] wherein, is the equipment health index, and the value range is 0 to 1; is an exponential function with a natural constant as the base, which is used to simulate the exponential attenuation law of the equipment performance; is an attenuation coefficient; is the equipment use time, and the unit is year; ​​is the failure frequency, unit: times per year; is the corrosion index; is the error weight coefficient; is the natural constant, approximately equal to 2.71828; is the error weight adjustment base, used to convert the error weight coefficient into the health degree correction factor.

[0118] The value of the attenuation coefficient is determined according to the aging law of the equipment in the marine environment: , wherein The maximum value reflects that the failure frequency has the most significant impact on the health degree.

[0119] The calculation of the expected maintenance cost is specifically represented as follows:

[0120] ;

[0121] In the formula, is the expected maintenance cost, unit: ten thousand yuan; is the basic maintenance cost, unit: ten thousand yuan; represents the negative 2 power of the equipment health degree index, that is, the square of the reciprocal, used to amplify the impact of the decline in health degree on the maintenance cost; is the failure frequency impact coefficient, indicating that the maintenance cost increases by 10% for each additional 1 time per year failure frequency; is a random error term, the value range of which is determined according to the historical statistics of maintenance cost fluctuations, generally 5% to 15% of the basic maintenance cost.

[0122] Among them, the parameter acquisition method is: The historical data statistics are adopted, including steps 1: collecting the maintenance records of the equipment in the past 3 years; step 2: calculating the average annual maintenance cost; and step 3: adjusting the current basic maintenance cost through the inflation rate. The equipment running time record is obtained. The failure log statistics are calculated.

[0123] The specific implementation of step S04 is to calculate the dynamic economic parameters based on the equipment health state and environmental factors. The calculation of the environmental wear coefficient is specifically represented as follows:

[0124] ;

[0125] In the formula, is the environmental wear coefficient; is the equipment running intensity; is the corrosion index; is the power index, indicating the square root operation, used to slow down the growth rate of the environmental factor influence; is a sine function, used to simulate the periodic influence of the environmental factor; Pi is the ratio of the circumference of a circle to its diameter, taking the value 3.14159; K is the reference loss coefficient; K is the main effect influence coefficient; K is the periodicity influence coefficient; K is the periodicity adjustment parameter, used to control the period length of the sine function.

[0126] The calculation of the dynamic depreciation rate is specifically represented as follows:

[0127] ;

[0128] In the formula, K is the dynamic depreciation rate; K is the actual workload, in hours; K is the design workload, in hours; K is the standard depreciation rate; K is the environmental loss coefficient; K is the health adjustment coefficient, used to adjust the depreciation rate according to the equipment health status; K is the equipment health index.

[0129] The calculation of the spare parts demand is specifically represented as follows:

[0130] ;

[0131] In the formula, K is the spare parts demand; K is the basic spare parts demand; K is a step function that outputs 1 when the input is greater than 0, otherwise outputs 0; K is the original value of the equipment, in ten thousand yuan; K is the fault frequency threshold, in times per year, corresponding to the case where the annual failure rate exceeds 10%; K is the maintenance cost proportion threshold, indicating that the maintenance cost accounts for 5% of the original value of the equipment; K is the error weight coefficient threshold; K is the fault impact adjustment coefficient, indicating that the spare parts demand increases by 20% when the fault frequency exceeds the threshold; K is the maintenance cost impact adjustment coefficient, indicating that the spare parts demand increases by 30% when the maintenance cost exceeds the threshold; K is the error impact adjustment coefficient, indicating that the spare parts demand increases by 15% when the error weight exceeds the threshold.

[0132] The parameter acquisition method is as follows: The inventory theory is used to calculate and obtain, including the following steps: 1. Statistics of historical spare parts consumption; 2. Calculation of the standard deviation of spare parts consumption; 3. Determination of safety stock based on service level. Obtained through equipment running time monitoring. From the equipment technical specification. Adopt industry standard value, generally take 0.05 to 0.15 according to different equipment type, usually take 0.10 for marine monitoring equipment.

[0133] The specific implementation of step S05 is to construct a comprehensive marine asset value assessment model to predict the value of the equipment. The calculation of the equipment value assessment result is specifically represented as follows:

[0134] ;

[0135] In the formula, is the equipment value assessment result, with the unit of ten thousand yuan; is the original value of the equipment; is a continuous multiplication symbol, representing the cumulative depreciation effect from the first year to the year; is the service life; is the dynamic depreciation rate in the year; is the error feature fusion function; is the error stability matrix; is the error change matrix; is the model prediction error term, the value range is determined according to the prediction accuracy during model training, generally 2% to 8% of the original value of the asset.

[0136] The error feature fusion function is defined as:

[0137] ;

[0138] In the formula, is the Frobenius norm; is the maximum possible norm value of the corresponding matrix, which is determined by the 95% quantile of the matrix norm in historical data statistics; is the benchmark value coefficient; is the stable error influence coefficient, representing the adjustment strength of the stable error on the equipment value; is the change error influence coefficient, representing the adjustment strength of the change error on the equipment value, which is smaller than the stable error coefficient to reflect that the influence of the change error is relatively small.

[0139] Wherein, the parameter acquisition method is: From the equipment procurement contract. Calculated through the marine asset value assessment model, which is trained by using deep learning method, the input is the equipment operation parameter and the error matrix, and the output is the value correction factor.

[0140] The specific implementation of step S06 is the same as the foregoing, which will not be described in detail here.

[0141] The specific implementation of step S07 is to realize automatic matching of accounting subjects and generation of financial voucher templates based on a business rule engine. The calculation of the financial voucher amount is specifically represented as follows:

[0142] ;

[0143] In the formula, is the financial voucher amount, with units of ten thousand yuan; is the basic business amount, with units of ten thousand yuan; is a step function; is an error weight coefficient; is an error adjustment coefficient, representing a 5% uncertainty adjustment to the voucher amount when the error weight exceeds a threshold value; is an amount adjustment error term, with a value range determined according to the financial processing accuracy requirement, generally 1% to 3% of the basic amount.

[0144] The basic business amount is determined according to different types of financial business events:

[0145] ;

[0146] In the formula, is the unit price of spare parts, with units of yuan; is the depreciation ratio, with a value range of 0.1 to 0.5.

[0147] The parameter acquisition method is as follows: Obtained from the supplier's quotation sheet. Determined according to the equipment health index and technical evaluation, including step 1: technical state evaluation of the equipment; step 2: combined with market price change analysis; step 3: determination of the depreciation ratio.

[0148] The specific implementation of step S08 is the same as described above and will not be described in detail here.

[0149] In this embodiment, it is necessary to note that the data collection frequency adjustment coefficient formula uses the principle of piecewise function to dynamically adjust the data collection frequency according to the product of the equipment running intensity and the corrosion index. The innovation of this formula is to couple the environmental corrosion and the equipment load state for analysis, automatically increasing the data collection density when the equipment is running at high load in a high corrosion environment. Compared with the traditional fixed frequency collection method, this method can obtain more detailed equipment state information at critical moments, while saving system resources at low risk periods, realizing intelligent optimization of data collection, and providing a more accurate data basis for subsequent error analysis and financial decision-making.

[0150] Floating change matrix The construction principle is based on time series analysis method, and the dynamic feature matrix is formed by calculating the change amount of monitoring parameters at continuous time. The effect of the matrix design is to mathematically express the time sequence fluctuation mode of the device measurement data. Compared with the traditional single-point measurement value analysis method, the floating change matrix can capture the dynamic evolution law of the measurement accuracy of the device, provide more rich feature information for identifying different types of measurement errors, and make the subsequent error classification and analysis more accurate and reliable.

[0151] Bayesian posterior probability calculation formula Using Bayesian inference principle, the influence probability of various errors on monitoring parameters is calculated by combining prior knowledge and observation evidence. The innovative effect of this method is that it can make full use of historical error statistical data as prior knowledge, combine the observed measurement change mode, and dynamically update the influence degree evaluation of various errors. Compared with the traditional error judgment method based on experience threshold, the Bayesian method provides a more scientific and quantitative error influence analysis framework, which significantly improves the accuracy and reliability of error identification.

[0152] Singular value decomposition method Using matrix decomposition principle, the complex comprehensive error relationship matrix is separated into systematic error and random error components. The effect of this mathematical model is that it can extract stable systematic error patterns and changing random error characteristics from mixed error signals. Compared with the traditional method of uniformly processing all errors, singular value decomposition realizes the accurate separation of error components, provides a technical foundation for targeted processing and compensation of different types of errors, and greatly improves the accuracy and depth of error analysis.

[0153] Gating weight function Using adaptive weight allocation principle based on confidence, the weight coefficients of each error component are dynamically adjusted according to the recognition confidence of different error types. The innovative effect of this function design is to realize the intelligent quantification of error influence degree. When the system error confidence is high, increase its weight to highlight the influence of systematic problems. When the random error confidence is high, reduce its weight to reduce the interference of accidental fluctuations. Compared with the traditional fixed weight error processing method, the gating weight function can dynamically adjust the analysis strategy according to the actual error type, significantly improving the accuracy and adaptability of error evaluation.

[0154] Device health index formula The influence of time aging, failure frequency, environmental corrosion and measurement error on the health state of equipment is comprehensively considered by adopting the principle of multi-factor exponential attenuation model. The effect of the model is that the real health state of the equipment under the marine environment can be comprehensively reflected, the exponential attenuation function accurately simulates the nonlinear attenuation law of the equipment performance with various factors, and the introduction of error weight coefficient enables the health degree evaluation to consider the influence of measurement uncertainty. Compared with the traditional health degree evaluation method based on single parameter or linear model, the formula provides more scientific and comprehensive equipment state evaluation, and lays a reliable foundation for subsequent maintenance decision and asset value evaluation.

[0155] Environmental wear coefficient formula The compound modeling principle combining the main effect and periodic effect is adopted, the square root function reflects the marginal diminishing effect of environmental factor influence, and the sine function simulates the periodic variation characteristics of the marine environment. The innovative effect of the formula is that the complex influence mechanism of the marine environment on the value loss of equipment can be accurately reflected, compared with the traditional method of adopting fixed depreciation rate, the environmental wear coefficient can dynamically adjust the value attenuation speed of equipment according to the actual operation intensity and corrosion degree, so that the depreciation calculation is more in line with the actual use of the marine monitoring equipment, and a scientific basis is provided for accurate financial accounting.

[0156] Equipment value evaluation formula The cumulative depreciation and error characteristic fusion comprehensive modeling principle is adopted, the cumulative effect of depreciation is embodied in the form of continuous multiplication, and the measurement uncertainty is brought into the value evaluation process by the error characteristic fusion function. The outstanding effect of the model is to realize accurate asset value evaluation based on the actual running state and measurement quality. Compared with the traditional simple depreciation method based on time or yield, the formula can comprehensively consider the actual health state of equipment, environmental influence and measurement error and the like, and the accuracy and reliability of asset value evaluation are significantly improved, and a more scientific quantitative tool is provided for asset management and financial decision of the marine monitoring platform.

[0157] It should be noted that the variables involved in the present application are explained in detail as shown in Tables 1-1 and 1-2.

[0158] Table 1-1 Variable explanation table

[0159]

[0160] Table 1-2 Variable explanation table

[0161]

[0162] For a better understanding and implementation of the present application, the following provides an embodiment 2 of a specific application scenario of the present application: a research team deploys a set of marine environment monitoring platform in a certain sea area, which is equipped with multiple marine monitoring equipment, including temperature sensor, salinometer, pressure sensor, dissolved oxygen meter and other key equipment. The main technical problem faced by the research team is how to accurately assess the running status and value change of the equipment in the marine environment, and automatically generate the corresponding financial vouchers to improve the accuracy and efficiency of asset management.

[0163] In the implementation process, the research team first establishes a distributed Internet of Things sensor network to collect real-time data from each monitoring device. Taking a key multi-parameter water quality monitor as an example, the basic data collection frequency of the device is set to every 10 minutes. Through real-time monitoring, the device running intensity is 0.75, and the corrosion index is 6.8. According to the data collection frequency adjustment coefficient calculation formula, Since the product value is between 2 and 6, the data collection frequency adjustment coefficient α is 1.0, and the original collection frequency remains unchanged.

[0164] In a 30-day continuous monitoring period, the research team collected the running status data of the device. The cumulative usage time of the device reached 18520 hours, and the fault frequency was 0.12 times per year. The main parameters measured by the device include seawater temperature, salinity, pressure and dissolved oxygen content. Through time series change characteristic analysis, a 96x4 dimensional floating change matrix is constructed, which reflects the change pattern of the four monitoring parameters at 96 time points.

[0165] The research team uses Bayesian model to analyze the floating change matrix and establishes the association between error types and monitoring parameters. According to historical data statistics, the prior probabilities of systematic error, random error, environmental interference error and drift error are set to 0.3, 0.4, 0.2 and 0.1 respectively. Through Bayesian inference calculation, a 4x4 dimensional comprehensive error relationship matrix is obtained. Using singular value decomposition method, the matrix is decomposed into error stability matrix and error change matrix, and the cumulative contribution rate of the first two singular values reaches 87%, so the stable error component dimension k=2 is determined.

[0166] The error change matrix is input into a pre-trained marine equipment error classifier, which is trained based on 5000 historical error samples. The classification results show that the system error confidence is 0.85, the random error confidence is 0.72, the environmental interference error confidence is 0.63, and the drift error confidence is 0.48. According to the calculation of the gating weight function, since the system error confidence exceeds 0.8, the corresponding error weight coefficient is set to 1.5; the random error confidence exceeds 0.7, and the weight coefficient is set to 0.8; the environmental interference error confidence exceeds 0.6, combined with the corrosive index 6.8, the weight coefficient is adjusted to 1.0+0.3×6.8 / 10=1.204. The final error weight coefficient is calculated as =1.32.

[0167] Based on the collected operating state data, the research team calculates the equipment health index. The equipment usage time =2.11 years (18520 hours), the failure frequency =0.12 times per year, the corrosive index =6.8, and the error weight coefficient =1.32. Using the equipment health index calculation formula, we get =0.73, indicating that the overall state of the equipment is good but there is a certain degree of performance degradation.

[0168] Combined with the equipment health index and maintenance history data, the research team predicts future maintenance needs. The basic maintenance cost According to the maintenance records of the past 3 years, it is 85,000 yuan. Through the expected maintenance cost calculation formula, we get =14.2 million yuan, indicating that as the equipment health decreases and the error influence increases, the maintenance cost increases significantly.

[0169] In terms of dynamic economic parameter calculation, the environmental wear coefficient calculation result is =1.26, reflecting the accelerated impact of the marine environment on the equipment value loss. The design workload of this equipment is 87600 hours, and the actual workload is 18520 hours. The standard depreciation rate is taken as 0.10. According to the dynamic depreciation rate calculation formula, the current dynamic depreciation rate =0.029.

[0170] According to the failure frequency, expected maintenance cost, and error weight coefficient, the research team determines the spare parts demand. The basic spare parts demand According to historical consumption statistics, it is 25 pieces. Since the failure frequency 0.12 exceeds the threshold value 0.1, the expected maintenance cost accounts for 14.2 / 280=0.051 of the original value of the equipment, which exceeds the threshold value 0.05, and the error weight coefficient 1.32 exceeds the threshold value 1.2, the spare parts demand is adjusted to =25x1.2x1.3x1.15=44.85≈45pieces.

[0171] As shown in Table 2, the equipment operation parameter statistical data details the changes in key indicators during the monitoring period.

[0172] Table 2 Equipment operation parameter statistical data

[0173]

[0174] The marine asset value evaluation model constructed by the research team uses 800 devices with more than 2 years of historical data for training. The model takes running intensity, corrosion index, failure frequency, expected maintenance cost, error stability matrix, and error change matrix as input parameters. After error feature fusion layer processing, combined with time series data analysis, the remaining useful life of the device is predicted to be 6.3 years. The original value of the device is 2.8 million yuan, and after the value evaluation model calculation, the current equipment value evaluation result =246.8 million yuan.

[0175] Based on the changes in the equipment health index, the expected maintenance cost, the equipment value evaluation result, and the error weight coefficient, the financial business event identification rule base automatically identifies three types of financial business events. First, the equipment depreciation event, as the current dynamic depreciation rate is 0.029, triggering the need for monthly depreciation voucher generation. Second, the maintenance cost event, the expected maintenance cost of 14.2 million yuan exceeds the budget threshold of 12 million yuan, and the error weight coefficient of 1.32 is greater than 1.0, triggering the generation of maintenance cost vouchers. Third, the spare parts procurement event, the demand for spare parts of 45 pieces exceeds the current inventory of 35 pieces, and the error weight coefficient shows that the risk of equipment failure increases, triggering the need for procurement vouchers.

[0176] As shown in Table 3, the financial business event identification result records the various financial processing needs automatically identified by the system.

[0177] Table 3 Financial business event identification result

[0178]

[0179] In the automatic matching process of accounting subjects, the system calls the pre-configured accounting subject mapping rules. The equipment depreciation event corresponds to the fixed asset depreciation subject, and the voucher amount is 246.8x0.029=7.15 million yuan. The maintenance cost event corresponds to the management expense subject, and since the error weight coefficient 1.32 exceeds 1.2, a 5% uncertainty adjustment is added to the base amount of 14.2 million yuan, resulting in a final voucher amount of 14.91 million yuan. The spare parts procurement event corresponds to the raw material procurement subject, with a spare part unit price of 620 yuan and a demand of 45 pieces, totaling 2.79 million yuan.

[0180] The generated financial voucher template is subjected to multi-level compliance verification by the rule engine. The verification content includes voucher element integrity verification, confirmation that all necessary accounting items, amounts and summary information have been filled in; debit-credit balance verification to ensure that the debit amount of each voucher is equal to the credit amount; amount reasonableness verification to confirm that the amount is within a reasonable range by comparing with historical data; and business logic consistency verification to ensure that the voucher content is correct in relation to the logic of the triggering business event.

[0181] As shown in Table 4, the compliance verification result shows the passing of each verification index.

[0182] Table 4 Compliance verification result

[0183]

[0184] After the verification passes, the system automatically enters the financial voucher into the enterprise financial management system. At the same time, a complete audit trail record is generated, recording the whole process information from data collection, error analysis, event identification to voucher generation. The audit trail contains the timestamp, operation content, related parameters and processing result of each processing link, ensuring the completeness and traceability of the financial processing process. The whole processing process takes 42 minutes, which greatly improves the efficiency compared with the 4 hours of manual processing.

[0185] To solve the core technical problem of asset management and financial processing of marine monitoring equipment, the traditional method mainly relies on manual regular inspection and subjective judgment for equipment state evaluation, uses fixed depreciation rate to calculate asset value, and judges maintenance demand and spare parts procurement opportunity by manual experience, and financial voucher generation completely depends on manual operation of accountants. The present application has made significant progress compared with the traditional method: in the accuracy of equipment state evaluation, the evaluation accuracy is improved from 76% of the traditional method to 89% through multi-dimensional error analysis; in the calculation of asset value, the dynamic depreciation rate reduces the value evaluation deviation from 18% to 6% compared with the fixed depreciation rate; in the prediction of maintenance cost, the prediction accuracy is improved from 68% of the traditional experience judgment to 83%; in the efficiency of financial processing, the automatic voucher generation shortens the processing time from 4 hours to 0.7 hours, and the processing accuracy is improved from 92% of manual operation to 98%. These technical progress makes the asset management of marine monitoring equipment more accurate and efficient, and provides more reliable technical support for marine scientific research and monitoring work.

[0186] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for generating financial vouchers based on a marine monitoring platform, characterized in that, include: By collecting real-time operational status data, environmental parameter data, and equipment measurement data of marine monitoring equipment through IoT sensors, an equipment operation database is established, while historical maintenance data is recorded and the data collection frequency is dynamically adjusted. Equipment measurement data is extracted from the equipment operation database and analyzed for time-series variation characteristics to construct a floating variation matrix. The floating variation matrix is ​​analyzed using a Bayesian model to obtain a comprehensive error relationship matrix. Singular value decomposition is performed on the comprehensive error relationship matrix to obtain an error stability matrix and an error variation matrix. The error variation matrix is ​​input into a marine equipment error classifier for classification to obtain error classification results. Based on the error classification results, error weight coefficients are calculated using a gating weight function. Based on the error classification results, the system performs status analysis and anomaly detection on the operational status data, calculates the equipment health index and expected maintenance costs, calculates the dynamic depreciation rate and spare parts demand, establishes a marine asset valuation model to calculate the equipment valuation results, automatically identifies financial and business events based on changes in the equipment health index, expected maintenance costs, equipment valuation results, and error weighting coefficients, and automatically matches the corresponding accounting subjects according to the accounting subject mapping rules based on the financial and business events, generates financial voucher templates, and enters them into the financial system. In the steps of calculating the dynamic depreciation rate and spare parts demand, the environmental loss coefficient is calculated based on the equipment health index and expected maintenance cost, combined with the operating intensity and corrosion index. The dynamic depreciation rate is calculated by combining the workload method with the environmental loss coefficient. The spare parts demand is determined based on the failure frequency, expected maintenance cost and error weighting coefficient. The adjustment coefficient of the data acquisition frequency is a frequency adjustment parameter calculated based on the operating intensity and corrosion index, which is used to dynamically adjust the data acquisition interval of the Internet of Things sensor. The marine equipment error classifier is an error type identification model built on deep learning technology. It is used to automatically identify and classify different types of equipment measurement errors, and the output error classification results include the probability distribution of each error type and the final classification label. The marine asset valuation model is an asset value prediction model built based on time series data analysis. It takes operating intensity, corrosion index, failure frequency, expected maintenance cost, error stability matrix and error change matrix as input parameters, processes the error stability matrix and error change matrix through error feature fusion layer, and predicts the remaining service life of the equipment by combining time series data analysis.

2. The method for generating financial vouchers based on a marine monitoring platform according to claim 1, characterized in that, The floating change matrix is ​​a dynamic change feature matrix constructed by mathematically modeling the time-series change characteristics of equipment measurement data, reflecting the fluctuation pattern of equipment measurement accuracy.

3. The method for generating financial vouchers based on a marine monitoring platform according to claim 2, characterized in that, The Bayesian model analysis process includes using the change patterns of measurement data in the fluctuation change matrix as observational evidence, establishing prior probability distributions of systematic errors, random errors, environmental interference errors, and drift errors, calculating the posterior probability of the influence of various errors on different monitoring parameters through Bayesian inference methods, and constructing a comprehensive error relationship matrix that includes the correlation strength between error type and monitoring parameter.

4. The method for generating financial vouchers based on a marine monitoring platform according to claim 3, characterized in that, The calculation method of the gating weight function is based on the confidence score of each error type in the error classification results. When the confidence score of the systematic error exceeds 0.8, the corresponding error weight coefficient is set to 1.

5. When the confidence score of the random error exceeds 0.7, the corresponding error weight coefficient is set to 0.

8. When the confidence score of the environmental interference error exceeds 0.6, the corresponding error weight coefficient is dynamically adjusted to between 1.0 and 1.3 according to the corrosion index.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions, which, when executed in a computer, are used to perform a method for generating financial vouchers based on a marine monitoring platform as described in any one of claims 1-4.

6. A financial voucher generation system based on a marine monitoring platform, characterized in that, The system includes the computer-readable storage medium of claim 5, wherein the system is any one of a computer, a server, or a microcontroller, the computer-readable storage medium is disposed within the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

Images