Ship integrated navigation system based on assistance of electromagnetic log

Abstract

The invention discloses a ship integrated navigation system based on assistance of an electromagnetic log, and relates to the technical field of ship integrated navigation, the ship integrated navigation system comprises a sensor unit, a data preprocessing unit, a double-layer fusion processing engine and a navigation output unit; the sensor unit comprises a strapdown inertial navigation system and an electromagnetic log and is used for collecting original inertial data and original speed data of a ship. According to the integrated ship navigation system based on assistance of the electromagnetic log, the quick response layer preferentially outputs low-delay navigation estimation through a simplified calculation model, and the real-time requirement is met; the precise arbitration layer guarantees the estimation precision through a more complete model, reduces the calculation burden, can start a closed-loop processing flow containing reason diagnosis and instructive backtracking when the estimation is detected to be inconsistent, corrects the deviation through targeted recalculation or state backtracking, and improves the accuracy of estimation. And the ability of the system to cope with complex conditions such as ocean current change and sensor instantaneous abnormality is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship integrated navigation, in particular to a ship integrated navigation system based on electromagnetic log auxiliary. BACKGROUND

[0002] Ship navigation technology is the key support to ensure the safety and efficiency of maritime navigation. The SINS (Strapdown Inertial Navigation System) is widely used in ship navigation due to its strong autonomy and good concealment. However, the positioning error of SINS will gradually diverge with time, so it is often combined with other external sensors for integrated navigation to improve accuracy and reliability. EM Log (Electromagnetic Log) is a common ship speed measuring device, which is usually used to provide speed information to assist SINS in error damping, forming a SINS / EM Log integrated navigation mode.

[0003] However, in the actual navigation environment, the electromagnetic log measures the speed of the ship relative to the ocean current, and is affected by the dynamic changes of the ocean current and the time-varying characteristics of the measurement noise in the complex marine environment. Its speed measurement is prone to significant errors, which in turn affects the overall positioning accuracy of the integrated navigation system. Therefore, how to effectively suppress the interference caused by ocean current and time-varying measurement noise in SINS / EM Log integrated navigation has become an important issue to improve the performance of ship navigation.

[0004] Currently, there have been some researches trying to solve the above problems. For example, the authorized publication "CN116164752A" discloses a ship integrated navigation method assisted by electromagnetic log. This method establishes a fixed model set containing two sub-models, and uses the expectation model expansion method to adaptively generate the model set. It combines the variational Bayesian filtering based on the mean field theory to decouple the state and noise covariance, and finally improves the positioning accuracy through multi-model fusion. This method improves the adaptability of the navigation system to ocean current and time-varying noise to some extent.

[0005] However, the existing method still has the following limitations: first, the model set construction and filtering process have high complexity, which poses a great challenge to the real-time processing capability of the ship navigation system; second, this method mainly focuses on algorithm optimization, and does not fully address the reliability issues in system integration design and practical engineering applications; in addition, in the multi-model filtering process, the model switching and fusion mechanism may still be limited by the prior parameter settings, affecting the robustness of the system in extreme marine environments.

[0006] Therefore, it is necessary to further develop an electromagnetic log assisted ship integrated navigation system with more optimized system structure, higher computational efficiency and stronger adaptability on the basis of existing technology, in order to better meet the comprehensive needs of accuracy, real-time performance and reliability in actual ship navigation. SUMMARY

[0007] The purpose of this invention is to provide a ship integrated navigation system based on an electromagnetic log to solve the problems mentioned in the background art.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a ship integrated navigation system based on electromagnetic log assist, comprising a sensor unit, a data preprocessing unit, a dual-layer fusion processing engine, and a navigation output unit; The sensor unit includes a strapdown inertial navigation system and an electromagnetic log, used to collect the ship's raw inertial data and raw speed data; The data preprocessing unit is connected to the sensor unit and is used to perform time synchronization and gross error removal on the raw inertial data and raw velocity data to form synchronized measurement data. The dual-layer fusion processing engine is connected to the data preprocessing unit and is used to perform parallel processing and fusion estimation on the synchronous measurement data. The dual-layer fusion processing engine includes a fast response layer and a precision arbitration layer set in parallel, an intermediate result cache connected between the fast response layer and the precision arbitration layer, and a dynamic feedback closed-loop controller connected to the fast response layer, the precision arbitration layer and the intermediate result cache respectively. The fast response layer is configured to perform a first fusion filtering process on the synchronous measurement data based on a simplified state vector and a simplified model set, generate a first navigation state estimate and a corresponding first covariance, and perform real-time verification on the first navigation state estimate based on preset rules. The precision arbitration layer is configured to perform a second fusion filtering process on the synchronous measurement data based on the complete state vector and the complete model set to generate a second navigation state estimate and the corresponding second covariance. The dynamic feedback closed-loop controller is configured to, after the immediate verification passes, selectively trigger a command backtracking operation based on the consistency determination result between the first navigation state estimate and the second navigation state estimate, and update the model parameters or state information stored in the intermediate result cache based on the backtracking result; The precision arbitration layer is also configured to recalculate the cached synchronization measurement data based on updated model parameters or state information after receiving a trigger command. The navigation output unit is connected to the dual-layer fusion processing engine and is used to correct the error of the strapdown inertial navigation system based on the final navigation state estimate output by the dual-layer fusion processing engine, and output the corrected navigation parameters.

[0009] Furthermore, the fast response layer includes a first filtering unit and an instant verification unit; The first filtering unit is used to run an interactive multi-model filtering algorithm based on a simplified state vector, wherein the simplified state vector includes at least position error, velocity error, and attitude misalignment angle; the simplified model set contains a finite number of models corresponding to typical ship motion states. The instant verification unit is connected to the first filtering unit and is used to perform instant verification on the first navigation state estimate; the rules for instant verification include physical limit check rules, short-term consistency check rules with the state estimate of the previous cycle, and inertial data residual change rate check rules.

[0010] Furthermore, the precision arbitration layer includes a second filtering unit and a multi-source consistency arbitration unit; The second filtering unit is used to run an interactive multi-model adaptive filtering algorithm based on the complete state vector; the complete state vector further includes gyroscope drift, accelerometer bias and ocean current velocity estimation based on the simplified state vector. The multi-source consistency arbitration unit is connected to the first filtering unit, the second filtering unit, and the instant verification unit, respectively. The multi-source consistency arbitration unit is configured to: calculate the state difference between the first navigation state estimate and the second navigation state estimate when the output of the instant verification unit is "pass" and the second filtering unit has completed the calculation; perform a consistency hypothesis test based on the first covariance, the second covariance, and the state difference; determine whether the first navigation state estimate and the second navigation state estimate are consistent based on the test result, and output an arbitration command.

[0011] Furthermore, the dynamic feedback closed-loop controller includes a state health assessment unit and a backtracking scheduling unit; The state health assessment unit is connected to the multi-source consistency arbitration unit, the first filtering unit, and the second filtering unit. The state health assessment unit is configured to: when the arbitration command indicates that the first navigation state estimate and the second navigation state estimate are inconsistent, receive model probability information from the first filtering unit, model probability information from the second filtering unit, and residual information, and analyze them in conjunction with the characteristic patterns of the state difference quantity to determine the potential cause categories of inconsistency; the potential cause categories include at least model mismatch categories, sensor instantaneous anomaly categories, and environmental mutation categories. The backtracking scheduling unit is connected to the state health assessment unit, the second filtering unit, and the intermediate result cache, respectively; the backtracking scheduling unit is configured to generate and send different backtracking control instructions based on the potential cause categories output by the state health assessment unit. When the potential cause category is model mismatch or environmental mutation, a first type of backtracking instruction is generated; the first type of backtracking instruction is used to instruct the second filtering unit to adjust its model set parameters or introduce a temporary model, and to recalculate based on the current and historical period synchronous measurement data stored in the intermediate result cache. When the potential cause category is a sensor transient anomaly category, a second type of backtracking instruction is generated; the second type of backtracking instruction is used to instruct the navigation system to roll back the system state to the most recent state point determined to be consistent by the multi-source consistency arbitration unit, and to instruct the first filtering unit to reinitialize based on the rolled-back state.

[0012] Furthermore, the multi-source consistency arbitration unit is also configured to, after the second filtering unit completes recalculation according to the backtracking control instruction, arbitrate again with the newly recalculated second navigation state estimate and the first navigation state estimate corresponding to the triggering backtracking or the new first navigation state estimate, to form a cyclic verification process.

[0013] Furthermore, the intermediate result cache is used to cyclically store the synchronization measurement data, the first navigation state estimate, the second navigation state estimate, and the consistency history determined by the multi-source consistency arbitration unit for a set time period.

[0014] Furthermore, the state health assessment unit is also configured to generate a long-term system health score based on the records of historical inconsistent events, the model probability evolution trend, and the residual sequence characteristics; the dynamic feedback closed-loop controller adaptively fine-tunes the process noise parameters or model transition probability parameters of the first filtering unit and the second filtering unit based on the long-term system health score.

[0015] Furthermore, the navigation output unit is configured to, when the multi-source consistency arbitration unit determines that the first navigation state estimate and the second navigation state estimate are consistent, preferentially select the first navigation state estimate as the final navigation state estimate.

[0016] Furthermore, the system also includes an auxiliary sensor unit, which includes at least one of a global navigation satellite system receiver or an underwater acoustic positioning device; when performing a consistency hypothesis test, the multi-source consistency arbitration unit also introduces the auxiliary positioning information provided by the auxiliary sensor unit at the corresponding time as a reference.

[0017] Furthermore, the system is integrated into the ship navigation control computer, and the dual-layer fusion processing engine is implemented as a software module on the processor of the navigation control computer. The operation of the dynamic feedback closed-loop controller is independent of the main filtering thread of the fast response layer and the precision arbitration layer.

[0018] This invention provides a ship integrated navigation system based on an electromagnetic log assisted by an electromagnetic log. It has the following beneficial effects: This electromagnetic log-assisted ship navigation system effectively improves the overall efficiency of navigation information processing by introducing a parallel processing dual-layer fusion architecture and a dynamic feedback closed-loop control mechanism. The fast response layer prioritizes low-latency navigation estimates using a simplified computational model to meet real-time requirements; the precision arbitration layer ensures estimation accuracy through a more complete model. The results from both layers are compared via a multi-source consistency arbitration unit. This design allows the system to prioritize the results from the fast channel during most normal navigation periods, reducing computational burden. When inconsistencies in estimations are detected, a closed-loop processing flow including cause diagnosis and command backtracking is initiated. Targeted recalculation or state rollback corrects deviations, enhancing the system's ability to handle complex situations such as ocean current changes and instantaneous sensor anomalies.

[0019] This electromagnetic log-assisted ship navigation system possesses adaptive adjustment capabilities based on long-term operational status. By recording and analyzing inconsistencies, the dynamic feedback closed-loop controller can slowly fine-tune internal filter parameters, allowing the system behavior to gradually adapt to the current navigation environment and sensor characteristics. This design reduces reliance on fixed prior parameters and improves long-term operational stability. Simultaneously, information from auxiliary sensors is incorporated into the consistency arbitration decision, further enhancing the confidence level of state estimation reliability verification. The entire solution is implemented in software through modular threads, ensuring the system's feasibility and maintainability in engineering applications, providing ships with a more robust and efficient integrated navigation solution. Attached Figure Description

[0020] Figure 1 This is a system module data flow diagram of a ship integrated navigation system based on an electromagnetic log assisted by the present invention; Figure 2 This is a backtracking decision tree diagram for a ship integrated navigation system based on an electromagnetic log assisted by the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 and Figure 2The present invention provides a technical solution: a ship integrated navigation system based on electromagnetic log assist, comprising a sensor unit, a data preprocessing unit, a dual-layer fusion processing engine, and a navigation output unit; The sensor unit includes a strapdown inertial navigation system and an electromagnetic log, used to collect the ship's raw inertial data and raw speed data; The data preprocessing unit, connected to the sensor unit, is used to perform time synchronization and gross error removal on the raw inertial data and raw velocity data to form synchronized measurement data; A dual-layer fusion processing engine, connected to the data preprocessing unit, is used to perform parallel processing and fusion estimation of synchronous measurement data. The dual-layer fusion processing engine includes a fast response layer and a precision arbitration layer set in parallel, an intermediate result cache connected between the fast response layer and the precision arbitration layer, and a dynamic feedback closed-loop controller connected to the fast response layer, the precision arbitration layer and the intermediate result cache respectively. The fast response layer is configured to perform a first fusion filtering process on the synchronous measurement data based on the simplified state vector and the simplified model set, generate a first navigation state estimate and the corresponding first covariance, and perform real-time verification on the first navigation state estimate based on preset rules. The precision arbitration layer is configured to perform a second fusion filtering process on the synchronous measurement data based on the complete state vector and the complete model set, to generate a second navigation state estimate and the corresponding second covariance. The dynamic feedback closed-loop controller is configured to selectively trigger a command backtracking operation based on the consistency determination result between the first navigation state estimate and the second navigation state estimate after the immediate verification is passed, and update the model parameters or state information stored in the intermediate result cache based on the backtracking result. The precision arbitration layer is also configured to recalculate the cached synchronous measurement data based on updated model parameters or state information after receiving a trigger command; The navigation output unit, connected to the dual-layer fusion processing engine, is used to correct errors in the strapdown inertial navigation system based on the final navigation state estimate output by the dual-layer fusion processing engine, and output the corrected navigation parameters.

[0023] It should be further explained that the specific implementation of the ship integrated navigation system based on electromagnetic log assistance is as follows: The system hardware includes a strapdown inertial navigation system, an electromagnetic log, and a navigation computer; the software is implemented on the navigation computer. The workflow is as follows: The angular velocity and specific force information output by the strapdown inertial navigation system, along with the water velocity information output by the electromagnetic log, first enter the data preprocessing unit for timestamp alignment, and outliers with obvious abnormalities are removed according to preset thresholds to form a synchronized and preliminarily purified synchronous measurement data packet. This data packet is then simultaneously sent to the fast response layer and the precision arbitration layer of the dual-layer fusion processing engine.

[0024] A simplified interactive multi-model filter runs within the fast response layer. Its state vector contains only three-dimensional position error, three-dimensional velocity error, and three-axis misalignment angle. The model set includes, for example, models describing uniform linear motion and models describing uniform turning motion. After processing the input data, the filter generates a first navigation state estimate and a corresponding first covariance matrix. Subsequently, an instant verification unit checks the estimate. The verification rules include determining whether the velocity value exceeds the physical limit of the ship's maximum speed, whether the difference between the current estimate and the previous cycle's filtered prediction is within a preset range, and whether the residual rate of change calculated using the difference between the inertial calculated velocity and the filtered estimated velocity is stable.

[0025] The precision arbitration layer operates in parallel with a more complete interactive multi-model adaptive filter. Its state vector, based on the fast response layer's state, adds gyroscope constant drift, accelerometer constant zero bias, and eastward and northward ocean current velocity components. The model set is also richer, potentially including additional models to describe gradually changing or abrupt ocean currents. This filter outputs a second navigation state estimate and the corresponding second covariance matrix. Once the fast response layer's immediate validation passes and the precision arbitration layer completes its calculations, intermediate results are cached to record relevant data for the current period.

[0026] The multi-source consistency arbitration unit in the dynamic feedback closed-loop controller begins operation. It calculates the difference between the first and second navigation state estimates in each state component and, combined with the two covariance matrices, performs a statistical hypothesis test by calculating the Mahalanobis distance to determine whether the two are consistent. If they are consistent, the navigation output unit adopts the first navigation state estimate as the final result to correct the strapdown inertial navigation system. If they are inconsistent, the state health assessment unit is activated. It analyzes the model probabilities, innovation sequences, and pattern characteristics of the inconsistency difference vector of the two filters to determine whether the root cause of the inconsistency is model parameter misfitting, instantaneous jumps in electromagnetic log data, or drastic changes in the ocean current environment.

[0027] Based on the evaluation results, the backtracking scheduling unit issues corresponding instructions: if it is determined that the model parameters are unsuitable or the environment has changed, it instructs the precision arbitration layer to adjust the noise parameters of its specific model or temporarily add a model, and uses the synchronous measurement data of the most recent few cycles stored in the intermediate result cache to re-execute the filtering calculation; if it is determined that the sensor data is suspected to be abnormal, it instructs the overall system state to revert to the most recently marked consistent state point, and instructs the fast response layer to re-initialize and run from this reverted state. After the instructional backtracking operation is completed, the new results generated will be submitted to the consistency arbitration unit for review again, forming a closed-loop verification.

[0028] Throughout the process, intermediate result cache continuously stores data, and the dynamic feedback closed-loop controller also slowly and adaptively fine-tunes certain process noise parameters of the two filters based on long-term consistency history and health scores, enabling the system to have a gradual self-adjustment capability. The navigation output unit finally converts the stable fusion result into attitude, velocity, and position correction information and feeds it back to the strapdown inertial navigation system, outputting navigation parameters for display.

[0029] The fast response layer includes a first filtering unit and an instant verification unit; The first filtering unit is used to run an interactive multi-model filtering algorithm based on simplified state vectors. The simplified state vectors include at least position error, velocity error, and attitude misalignment angle. The simplified model set contains a finite number of models corresponding to typical ship motion states. The instant verification unit, connected to the first filtering unit, is used to perform instant verification on the first navigation state estimate. The rules for instant verification include physical limit check rules, short-term consistency check rules with the state estimate of the previous cycle, and inertial data residual change rate check rules.

[0030] Further explanation is needed regarding the specific implementation details of the fast response layer: The simplified state vector used by the first filtering unit of this layer is a nine-dimensional vector, specifically composed of seven basic navigation error states: attitude misalignment angles in the east, north, and sky directions; velocity errors in the east and north directions; and latitude and longitude errors. In its simplified interactive multi-model filtering algorithm, the model set specifically includes two sub-models. The first sub-model is a uniform linear motion model, in which the terms related to angular velocity in the state transition matrix are set to zero. The second sub-model is a uniform turning motion model, in which fixed parameters based on the typical turning angular velocity of the ship are introduced into the state transition matrix.

[0031] The first filtering unit performs recursive filtering calculations on the received synchronous measurement data, outputting a first navigation state estimate containing the aforementioned nine state estimates and its corresponding ninth-order first covariance matrix. The instantaneous verification unit is directly connected to the data output of the first filtering unit. It receives the first navigation state estimate and performs the following serialization rule checks: First, a physical limit check is performed to determine whether the speed estimate exceeds the maximum speed threshold preset according to the ship type, and whether the position change exceeds the possible displacement range calculated based on the position and maximum speed of the previous cycle; then, a short-time consistency check is performed, calculating the difference between the position and speed components in the current cycle's first navigation state estimate and the one-step prediction value of the previous cycle within the first filtering unit, and determining whether the norm of this difference vector is less than the dynamic threshold derived from the first covariance matrix. Finally, an inertial data residual change rate check is performed. The velocity residual for the current cycle is calculated using the difference between the velocity calculated by the strapdown inertial navigation system and the velocity corrected by the first navigation state estimation. The absolute value of the difference between this residual and the residual from the previous cycle is also calculated. If this absolute value continuously exceeds a set threshold, an anomaly is flagged. After all check rules are passed, the instant verification unit outputs a "pass" flag, which, along with the first navigation state estimation and the first covariance matrix, is output to subsequent units.

[0032] The precision arbitration layer includes a second filtering unit and a multi-source consistency arbitration unit; The second filtering unit is used to run an interactive multi-model adaptive filtering algorithm based on the complete state vector; the complete state vector further includes gyroscope drift, accelerometer bias and ocean current velocity estimation on the basis of the simplified state vector. A multi-source consistency arbitration unit is connected to a first filtering unit, a second filtering unit, and an instant verification unit, respectively. The multi-source consistency arbitration unit is configured to: when the output of the instant verification unit is "pass" and the second filtering unit has completed its calculation, calculate the state difference between the first navigation state estimate and the second navigation state estimate; perform a consistency hypothesis test based on the first covariance, the second covariance, and the state difference; and determine whether the first navigation state estimate and the second navigation state estimate are consistent based on the test results, and output an arbitration command.

[0033] Further explanation is needed regarding the specific implementation details of the precision arbitration layer: The second filtering unit of this layer uses a 17-dimensional state vector, which, based on the 9-dimensional simplified state vector, extends and adds constant drift of the three-axis gyroscope, constant zero bias of the three-axis accelerometer, and ocean current velocity components in the east and north directions; it runs a complete interactive multi-model adaptive filtering algorithm, with a model set containing more than two sub-models. In addition to covering uniform linear and turning models, it also includes a slowly varying ocean current model and a sudden ocean current model. The slowly varying ocean current model sets the process noise covariance of the ocean current state to a smaller value, while the sudden ocean current model sets it to a larger value. It also uses variational Bayesian inference to jointly estimate the statistical characteristics of state and measurement noise online.

[0034] The multi-source consistency arbitration unit receives in real time the first navigation state estimate and the first covariance matrix marked with a "pass" flag from the fast response layer, and the second navigation state estimate and the second covariance matrix from the second filtering unit. This unit first calculates the difference between the two state estimate vectors at their corresponding components to form a state difference vector. Then, based on this difference vector and the sum of the first and second covariance matrices, it constructs a scalar test statistic that follows a specific probability distribution. The arbitration unit compares this test statistic with a threshold value obtained from a table based on a preset confidence level. If the statistic is below the threshold, the first and second navigation state estimates are considered consistent, and an instruction is output to have the navigation output unit adopt the first navigation state estimate. If the statistic is above or equal to the threshold, the two are considered inconsistent, and a trigger signal is immediately sent to the dynamic feedback closed-loop controller to initiate the subsequent health assessment and backtracking process.

[0035] The dynamic feedback closed-loop controller includes a state health assessment unit and a backtracking scheduling unit; The state health assessment unit is connected to the multi-source consistency arbitration unit, the first filtering unit, and the second filtering unit. The state health assessment unit is configured to: when the arbitration command indicates that the first navigation state estimate and the second navigation state estimate are inconsistent, receive model probability information from the first filtering unit, model probability information from the second filtering unit, and residual information, and analyze them in conjunction with the characteristic patterns of the state difference quantity to determine the potential cause categories of inconsistency; the potential cause categories include at least model mismatch categories, sensor instantaneous anomaly categories, and environmental mutation categories. The backtracking scheduling unit is connected to the state health assessment unit, the second filtering unit, and the intermediate result cache, respectively. The backtracking scheduling unit is configured to generate and send different backtracking control commands based on the potential cause categories output by the state health assessment unit. When the potential cause category is model mismatch or environmental mutation, a first type of backtracking instruction is generated. The first type of backtracking instruction is used to instruct the second filtering unit to adjust its model set parameters or introduce a temporary model, and to recalculate based on the current and historical synchronous measurement data stored in the intermediate result cache. When the potential cause category is the sensor instantaneous anomaly category, a second type of backtracking instruction is generated; the second type of backtracking instruction is used to instruct the navigation system to roll back the system state to the most recent state point determined to be consistent by the multi-source consistency arbitration unit, and to instruct the first filtering unit to reinitialize based on the rolled-back state.

[0036] Further explanation is needed regarding the specific implementation of the dynamic feedback closed-loop controller as follows: When the multi-source consistency arbitration unit determines inconsistency and sends a trigger signal, the state health assessment unit immediately initiates the analysis process. This unit simultaneously receives the current simplified model probabilities from the first filtering unit, the current complete model probabilities from the second filtering unit, and the filter innovation sequence, and combines them with the state difference vector calculated by the multi-source consistency arbitration unit. The assessment unit first checks whether the probability of the ocean current mutation model in the second filter exceeds its slowly changing model probability and persists for several cycles. At the same time, it analyzes whether the eastward and northward velocity difference components constitute a specific vector pattern related to the current heading. If the conditions are met, it is classified as an environmental mutation category. Secondly, the assessment unit calculates the recent variance of the second filter innovation sequence and compares it with the normal range statistically determined based on historical data. If the variance suddenly increases while the simplified model probability of the first filter does not change drastically, it is classified as a sensor instantaneous anomaly category. If the above two characteristics are not obvious, but the norm of the state difference vector persists in multiple state components and the model probability drifts slowly, it is classified as a model mismatch category.

[0037] The backtracking scheduling unit performs corresponding operations based on the received category determination results: For the environmental mutation or model mismatch category, the scheduling unit sends a first type of backtracking instruction to the second filtering unit. This instruction contains specific parameter adjustment content, such as instructing it to adjust the diagonal element value of the process noise covariance of the ocean current state to a specified multiple of the original value, or temporarily introduce an additional model with a specific initial probability. Subsequently, the second filtering unit extracts the synchronous measurement data sequence of the current and previous several cycles from the intermediate result cache, and re-executes the recursive filtering calculation from the earliest cached data to the current time using the adjusted model set and parameters. For the sensor instantaneous anomaly category, the scheduling unit sends a second type of backtracking instruction to the system. This instruction first controls the navigation output unit to stop using the current output, then queries and calls the complete system state estimate corresponding to the most recent time when it was marked as "consistent" by the multi-source consistency arbitration unit from the intermediate result cache. This value is used as the global system state for backtracking overlay. At the same time, a re-initialization command is sent to the first filtering unit to load the corresponding navigation component in the backtracking state into its internal state register, and to continue the subsequent filtering process from this starting point.

[0038] The multi-source consistency arbitration unit is also configured to, after the second filtering unit completes the recalculation according to the backtracking control command, arbitrate again with the new second navigation state estimate obtained by recalculation and the first navigation state estimate corresponding to the triggering backtracking or the new first navigation state estimate, to form a cyclic verification process.

[0039] Further explanation is needed regarding the specific implementation of the re-arbitration loop: After the second filtering unit completes the specified recalculation or system rollback and re-initialization operation based on the first or second type of instructions issued by the backtracking scheduling unit, the multi-source consistency arbitration unit automatically starts a new round of arbitration process; for cases where the first type of backtracking instruction has been executed, the arbitration unit reads the first navigation state estimate and its first covariance matrix corresponding to the triggering of this backtracking from the intermediate result cache, compares and judges it again with the new second navigation state estimate and its new second covariance matrix newly calculated and output by the second filtering unit according to the original hypothesis testing method; In the case of executing the second type of backtracking instruction, the arbitration unit waits for the first filtering unit of the fast response layer to reinitialize in the backtracking state and run stably for at least one filtering cycle before obtaining its newly generated first navigation state estimate and first covariance matrix. Then, it compares and judges them again with the second navigation state estimate and second covariance matrix output by the second filtering unit at the current time. The result of this second determination will supersede the previous one: if the determination is consistent, the system resumes its normal output process, and the navigation output unit selects and adopts the corresponding state estimate based on the new consistency conclusion; if the determination is still inconsistent, the result will serve as a new trigger signal fed back to the dynamic feedback closed-loop controller. The state health assessment unit will conduct a more in-depth analysis based on the inconsistency characteristics and historical records. The backtracking scheduling unit may then adopt different subsequent strategies, such as further adjusting more parameters for a second backtracking calculation under the model mismatch category, or triggering a diagnostic procedure for a specified sensor and temporarily reducing its weight in the fusion under the sensor anomaly category. This process ensures that the corrective measures taken for the initial inconsistency can be independently verified, forming a closed-loop control logic with a verification step.

[0040] The intermediate result cache is used to cyclically store synchronous measurement data, first navigation state estimate, second navigation state estimate, and consistency history determined by the multi-source consistency arbitration unit for a set time period.

[0041] Further explanation is needed regarding the specific implementation details of the intermediate result cache: This cache module allocates a circular storage area in the navigation computer's memory, designed to store specific data within multiple consecutive navigation cycles. Each storage cycle corresponds to a complete data record structure, which includes the following fields: a timestamp field, used to record the precise time corresponding to the data; a synchronous measurement data field, used to store the raw observations of the strapdown inertial navigation system and electromagnetic logger for that cycle after processing by the preprocessing unit; a first navigation state estimation field, used to store the state vector estimate output by the fast response layer for that cycle; a first covariance matrix field, used to store the corresponding covariance matrix; a second navigation state estimation field, used to store the state vector estimate output by the precision arbitration layer for that cycle; a second covariance matrix field, used to store the corresponding covariance matrix; and a consistency flag field, used to store the determination result of the consistency between the two state estimates for that cycle by the multi-source consistency arbitration unit.

[0042] The caching module adopts a first-in-first-out queue management mechanism. When a new navigation cycle ends, the system will generate a new record containing all the above fields and write it to the tail of the queue. At the same time, if the queue is full, it will automatically overwrite the oldest record at the head of the queue. In this way, the cache always retains the complete data context of the most recent set time period, which is five navigation cycles.

[0043] When a dynamic feedback closed-loop controller needs to perform a commanded backtracking operation, it can accurately retrieve all relevant field data from this circular queue based on the timestamp. For example, when executing a first type of backtracking command that requires recalculation, the backtracking scheduling unit can extract all synchronous measurement data records from the current time back to the previously specified time from the cache and submit them to the second filtering unit for further filtering.

[0044] The health assessment unit is also configured to generate a long-term health score for the system based on records of historical inconsistent events, model probability evolution trends, and residual sequence characteristics; the dynamic feedback closed-loop controller adaptively fine-tunes the process noise parameters or model transition probability parameters of the first and second filtering units based on the long-term health score of the system.

[0045] Further explanation is needed regarding the specific implementation details of the adaptive parameter fine-tuning function: After completing the analysis of inconsistent events and driving the backtracking operation, the status health assessment unit will also store the category of this event, the model information involved, the residual characteristics, and the type and result of the backtracking operation as an event record in a dedicated historical log. At the same time, this unit maintains a long-term system health score, which is a value that varies between zero and one, with an initial value of one. The score update logic is as follows: whenever an inconsistent event occurs and is classified as a model mismatch or environmental mutation, the score is multiplied by a decay factor slightly less than one. Whenever the system runs continuously for more than a set long period without an inconsistent event, the score is multiplied by a recovery factor slightly greater than one, but the upper limit of the score does not exceed one.

[0046] The dynamic feedback closed-loop controller periodically reads the health score. When the score falls below a set first threshold, the controller generates a parameter fine-tuning instruction. This instruction first acts on the second filtering unit, specifically by moderately increasing the diagonal elements of the corresponding process noise covariance matrix based on the specific state components frequently involved in recent event records in the historical logs. For example, it increases the process noise variance of the eastward ocean current velocity component by a fixed small increment. Simultaneously, the controller also acts on the first filtering unit, adjusting the elements in its internal model transition probability matrix. For example, it increases the probability of switching from a turning model back to a straight-line model, making the filter more inclined to return to a conservative motion model when encountering uncertainty. The magnitude of these parameter adjustments is proportional to the degree to which the health score deviates from the threshold, but there is an upper limit to the absolute value of each adjustment to prevent excessive changes. After parameter adjustments, the controller resets the event logs related to the adjustment and observes the results of consistency arbitration over several subsequent periods. If the score gradually recovers due to the absence of new events, the controller will no longer actively make further adjustments. The system thus achieves adaptive slow tuning based on long-term operating performance.

[0047] The navigation output unit is configured such that when the multi-source consistency arbitration unit determines that the first navigation state estimate and the second navigation state estimate are consistent, the first navigation state estimate is selected as the final navigation state estimate.

[0048] Further explanation is needed regarding the specific implementation of the navigation output unit: This unit monitors the arbitration command and consistency judgment flag output by the multi-source consistency arbitration unit in real time. When a judgment result with the flag "consistent" is received, the output logic control circuit immediately activates, selecting and reading the first navigation state estimate generated by the fast response layer from the data channel as the final navigation state estimate to be output. Subsequently, the internal calculation module of this unit calculates an attitude correction amount in the form of a direction cosine matrix based on the misalignment component in this final estimate, and sends it to the error correction input of the strapdown inertial navigation system through the data interface for feedback correction of its real-time calculated attitude. At the same time, the calculation module directly uses the velocity error and position error components in the final estimate as corrections. The values ​​are added to the current output speed and position of the strapdown inertial navigation system to obtain the corrected speed and position information. After completing the above calculations and feedback, the navigation output unit packages the corrected attitude, speed, and position information, along with the system's current health status code, into a standard format navigation data frame and sends it to the ship's display console and other systems that require navigation information via the communication bus. Throughout the process, if the multi-source consistency arbitration unit outputs a "disagreement" flag or is in the process of instructional backtracking, the navigation output unit will continue to use the last valid final navigation state estimate for output, or switch to using the second navigation state estimate for output according to a specific instruction from the dynamic feedback closed-loop controller, until the system receives a new "consistency" determination flag.

[0049] The system also includes an auxiliary sensor unit, which includes at least one of a global satellite navigation system receiver or an underwater acoustic positioning device; when performing consistency hypothesis testing, the multi-source consistency arbitration unit also introduces the auxiliary positioning information provided by the auxiliary sensor unit at the corresponding time as a reference.

[0050] It should be further explained that the specific implementation of the enhanced arbitration function of the auxiliary sensor unit is as follows: The global satellite navigation system receiver in the auxiliary sensor unit is connected to the navigation computer through a serial data bus, continuously receiving and processing satellite signals to provide latitude, longitude and speed information, while the underwater acoustic positioning device provides position coordinates relative to the seabed beacon periodically at a predetermined working depth through a dedicated interface; when the data from these auxiliary sensors is sent to the data preprocessing unit, it is timestamped together with the data from the main sensor and its data validity status is marked.

[0051] When the multi-source consistency arbitration unit performs the consistency hypothesis test, it first checks whether the auxiliary sensor data at the corresponding time is valid. If the global navigation satellite system (GNSS) data is valid, the arbitration unit calculates the difference between the horizontal position output by the GNSS at that time and the horizontal position in the first navigation state estimate, as well as the difference between the horizontal position and the horizontal position in the second navigation state estimate, forming two additional position difference scalars. These two scalars, along with the horizontal position difference component extracted from the state difference vector, are input into an extended test statistic calculation function. This function obtains the final consistency determination result by weighted fusion of these difference information. The weight of the GNSS position difference is dynamically set according to its positioning accuracy factor. If the underwater acoustic positioning data is valid but the GNSS is invalid, the arbitration unit uses a similar method, but calculates the difference between the underwater acoustic positioning horizontal position and the two state estimates.

[0052] This process introduces an external observation reference independent of inertial sensors and electromagnetic logs for arbitration decisions, improving the confidence and reliability of consistent judgment of the outputs of two parallel filters in complex environments.

[0053] The system is integrated into the ship's navigation and control computer. The dual-layer fusion processing engine is implemented as a software module on the processor of the navigation and control computer. The operation of the dynamic feedback closed-loop controller is independent of the main filtering thread of the fast response layer and the precision arbitration layer.

[0054] It should be further explained that the specific implementation details of system integration and software implementation are as follows: The system is deployed as a whole in the ship's navigation and control computer, which uses an industrial-grade processor with multi-core processing capabilities and runs a real-time operating system; all functional modules of the dual-layer fusion processing engine, including the first filtering unit and instant verification unit of the fast response layer, and the second filtering unit and multi-source consistency arbitration unit of the precision arbitration layer, are implemented in the form of independent software threads running on the real-time operating system, and each thread is assigned a fixed execution cycle and priority; The state health assessment unit and backtracking scheduling unit of the dynamic feedback closed-loop controller are implemented as an independent monitoring and control thread. The scheduling priority of this thread is set to be higher than the aforementioned filtering thread but lower than the system emergency interrupt. It receives event trigger messages from the multi-source consistency arbitration unit through the operating system's message queue mechanism and accesses the data cached in the intermediate result through the shared memory area. The intermediate result cache exists as a global memory area shared by all relevant threads, and its access is synchronized through a semaphore mechanism to ensure data consistency. The navigation output unit is implemented as a separate communication and output thread, responsible for periodically collecting the final determined state information and sending it in a formatted manner. The software threads exchange states, instructions and data through the inter-process communication mechanisms provided by the real-time operating system, including but not limited to message queues, shared memory and event flags, so as to collaboratively complete the complete navigation information processing flow from data acquisition, parallel processing, consistency arbitration, feedback control to result output on a single computing platform.

[0055] This system effectively improves the overall efficiency of navigation information processing by introducing a parallel processing dual-layer fusion architecture and a dynamic feedback closed-loop control mechanism. The fast response layer prioritizes low-latency navigation estimates using a simplified computational model to meet real-time requirements; the precision arbitration layer ensures estimation accuracy through a more complete model. The results from both layers are compared via a multi-source consistency arbitration unit. This design allows the system to prioritize the results from the fast channel during most normal navigation periods, thereby reducing the computational burden. When inconsistencies in estimation are detected, a closed-loop processing flow including cause diagnosis and command backtracking can be initiated to correct deviations through targeted recalculation or state rollback, enhancing the system's ability to cope with complex situations such as ocean current changes and instantaneous sensor anomalies.

[0056] Furthermore, the system possesses adaptive adjustment capabilities based on long-term operating conditions. By recording and analyzing inconsistencies, the dynamic feedback closed-loop controller can slowly fine-tune internal filter parameters, allowing the system behavior to gradually adapt to the current navigation environment and sensor characteristics. This design reduces reliance on fixed prior parameters and improves long-term operational stability. Simultaneously, information from auxiliary sensors is incorporated into the consistency arbitration decision, further enhancing the confidence level of state estimation reliability verification. The entire solution is implemented at the software level through modular threads, ensuring the system's feasibility and maintainability in engineering applications, providing ships with a more robust and efficient integrated navigation solution.

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0058] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A ship integrated navigation system based on an electromagnetic log assisted by an electromagnetic log, characterized in that, It includes a sensor unit, a data preprocessing unit, a dual-layer fusion processing engine, and a navigation output unit; The sensor unit includes a strapdown inertial navigation system and an electromagnetic log, used to collect the ship's raw inertial data and raw speed data; The data preprocessing unit is connected to the sensor unit and is used to perform time synchronization and gross error removal on the raw inertial data and raw velocity data to form synchronized measurement data. The dual-layer fusion processing engine is connected to the data preprocessing unit and is used to perform parallel processing and fusion estimation on the synchronous measurement data. The dual-layer fusion processing engine includes a fast response layer and a precision arbitration layer set in parallel, an intermediate result cache connected between the fast response layer and the precision arbitration layer, and a dynamic feedback closed-loop controller connected to the fast response layer, the precision arbitration layer and the intermediate result cache respectively. The fast response layer is configured to perform a first fusion filtering process on the synchronous measurement data based on a simplified state vector and a simplified model set, generate a first navigation state estimate and a corresponding first covariance, and perform real-time verification on the first navigation state estimate based on preset rules. The precision arbitration layer is configured to perform a second fusion filtering process on the synchronous measurement data based on the complete state vector and the complete model set to generate a second navigation state estimate and the corresponding second covariance. The dynamic feedback closed-loop controller is configured to, after the immediate verification passes, selectively trigger a command backtracking operation based on the consistency determination result between the first navigation state estimate and the second navigation state estimate, and update the model parameters or state information stored in the intermediate result cache based on the backtracking result; The precision arbitration layer is also configured to recalculate the cached synchronization measurement data based on updated model parameters or state information after receiving a trigger command. The navigation output unit is connected to the dual-layer fusion processing engine and is used to correct the error of the strapdown inertial navigation system based on the final navigation state estimate output by the dual-layer fusion processing engine, and output the corrected navigation parameters.

2. The ship integrated navigation system based on electromagnetic log assisted according to claim 1, characterized in that: The fast response layer includes a first filtering unit and an instant verification unit; The first filtering unit is used to run an interactive multi-model filtering algorithm based on a simplified state vector, wherein the simplified state vector includes at least position error, velocity error, and attitude misalignment angle; the simplified model set contains a finite number of models corresponding to typical ship motion states. The real-time verification unit is connected to the first filtering unit and is used to perform real-time verification on the first navigation state estimate. The rules for real-time verification include physical limit check rules, short-term consistency check rules with the state estimate of the previous cycle, and inertial data residual change rate check rules.

3. A ship integrated navigation system based on an electromagnetic log assisted by claim 2, characterized in that: The precision arbitration layer includes a second filtering unit and a multi-source consistency arbitration unit; The second filtering unit is used to run an interactive multi-model adaptive filtering algorithm based on the complete state vector; The complete state vector, based on the simplified state vector, further includes gyroscope drift, accelerometer zero bias, and ocean current velocity estimation; The multi-source consistency arbitration unit is connected to the first filtering unit, the second filtering unit, and the instant verification unit, respectively. The multi-source consistency arbitration unit is configured to: when the output of the instant verification unit is pass and the second filtering unit completes the calculation, calculate the state difference between the first navigation state estimate and the second navigation state estimate; and perform a consistency hypothesis test based on the first covariance, the second covariance, and the state difference. Based on the test results, determine whether the first navigation state estimate and the second navigation state estimate are consistent, and output an arbitration command.

4. A ship integrated navigation system based on an electromagnetic log assisted by claim 3, characterized in that: The dynamic feedback closed-loop controller includes a state health assessment unit and a backtracking scheduling unit; The health status assessment unit is connected to the multi-source consistency arbitration unit, the first filtering unit, and the second filtering unit. The state health assessment unit is configured to: when the arbitration instruction indicates that the first navigation state estimate and the second navigation state estimate are inconsistent, receive model probability information from the first filtering unit, model probability information from the second filtering unit, and residual information, and analyze them in conjunction with the characteristic patterns of the state difference quantity to determine the potential cause category of the inconsistency. The potential cause categories include at least model mismatch, sensor transient anomaly, and environmental mutation; The backtracking scheduling unit is connected to the state health assessment unit, the second filtering unit, and the intermediate result cache, respectively; the backtracking scheduling unit is configured to generate and send different backtracking control instructions based on the potential cause categories output by the state health assessment unit. When the potential cause category is model mismatch or environmental mutation, a first type of backtracking instruction is generated; the first type of backtracking instruction is used to instruct the second filtering unit to adjust its model set parameters or introduce a temporary model, and to recalculate based on the current and historical period synchronous measurement data stored in the intermediate result cache. When the potential cause category is a sensor transient anomaly category, a second type of backtracking instruction is generated; the second type of backtracking instruction is used to instruct the navigation system to roll back the system state to the most recent state point determined to be consistent by the multi-source consistency arbitration unit, and to instruct the first filtering unit to reinitialize based on the rolled-back state.

5. A ship integrated navigation system based on an electromagnetic log assisted by claim 4, characterized in that: The multi-source consistency arbitration unit is further configured to, after the second filtering unit completes recalculation according to the backtracking control command, arbitrate again with the newly recalculated second navigation state estimate and the first navigation state estimate corresponding to the triggering backtracking or the new first navigation state estimate, to form a cyclic verification process.

6. A ship integrated navigation system based on an electromagnetic log assisted by claim 5, characterized in that: The intermediate result cache is used to cyclically store the synchronization measurement data, the first navigation state estimate, the second navigation state estimate, and the consistency history determined by the multi-source consistency arbitration unit for a set time period.

7. A ship integrated navigation system based on an electromagnetic log assisted by claim 6, characterized in that: The state health assessment unit is also configured to generate a long-term system health score based on the record of historical inconsistent events, the probability evolution trend of the model, and the characteristics of the residual sequence. The dynamic feedback closed-loop controller adaptively fine-tunes the process noise parameters or model transition probability parameters of the first and second filtering units based on the system's long-term health score.

8. A ship integrated navigation system based on an electromagnetic log assisted by claim 7, characterized in that: The navigation output unit is configured to, when the multi-source consistency arbitration unit determines that the first navigation state estimate and the second navigation state estimate are consistent, preferentially select the first navigation state estimate as the final navigation state estimate.

9. A ship integrated navigation system based on an electromagnetic log assisted by claim 8, characterized in that: The system also includes an auxiliary sensor unit, which includes at least one of a global satellite navigation system receiver or an underwater acoustic positioning device; when performing a consistency hypothesis test, the multi-source consistency arbitration unit also introduces the auxiliary positioning information provided by the auxiliary sensor unit at the corresponding time as a reference.

10. A ship integrated navigation system based on an electromagnetic log assisted by claim 9, characterized in that: The system is integrated into the ship navigation control computer. The dual-layer fusion processing engine is implemented as a software module on the processor of the navigation control computer. The operation of the dynamic feedback closed-loop controller is independent of the main filtering thread of the fast response layer and the precision arbitration layer.

Images