Amphibious robot power system fault prediction method based on land and water working condition difference

Abstract

The invention discloses an amphibious robot power system fault prediction method based on land and water working condition difference, and belongs to the technical field of robot control systems, and the method comprises the steps: sensing land and water physical field characteristics in real time, and emerging an environment coupling topology state; monitoring an energy and information exchange flow, and obtaining a flow mode divergence as a fault precursor; receiving land and water environment switching identification, inducing relaxation and tending to a new convergence steady state; taking the new steady state as an initial condition, taking the flow mode divergence and the energy dissipation rate as evolution driving potential, and generating a minimum action quantity track; and driving parameter or weight adjustment along a minimum action quantity track to realize structural adaptation and toughness internalization. According to the method, a control strategy integrating dynamic topology modeling, prospective risk assessment and structural learning adaptation is adopted, the active fault prediction capacity and the self-adaptive stability in the land and water environment switching process can be improved, and the overall reliability and toughness of the robot are enhanced.

Description

Technical Field

[0001] This invention relates to the field of robot control system technology, and in particular to a method for predicting the failure of amphibious robot power systems based on differences in land and water conditions. Background Technology

[0002] As a special type of robot capable of performing tasks in two drastically different environments—water and land—amphibious robots rely heavily on their propulsion system for mobility and operational capabilities. The control system manages energy generation, transmission, and consumption, and coordinates the work of each drive unit. Especially during the transition from water to land or vice versa, the physical constraints such as buoyancy, drag, and friction change drastically, placing extremely high demands on the real-time performance, stability, and adaptability of the propulsion system control.

[0003] Existing methods for fault detection and control of amphibious robot power systems typically employ control strategies based on preset models and fault diagnosis techniques based on fixed thresholds. These methods often rely on static mathematical models and determine whether the system is in normal working order by monitoring readings of individual key sensors, such as motor current, voltage, or temperature. When sensor readings exceed preset safety thresholds, alarms are triggered or protective measures such as emergency shutdowns are implemented. When dealing with environmental transitions, some methods employ segmented control strategies, designing different sets of control parameters for water and land environments and abruptly switching parameters at the transition points.

[0004] However, the aforementioned existing technologies have significant technical shortcomings in practical applications. First, diagnostic methods based on static models and fixed thresholds are poorly adaptable to dynamically changing environments, making it difficult to capture early fault precursors accumulated due to gradual changes in operating conditions, often resulting in prediction lag or a large number of false alarms. Second, segmented control strategies are prone to oscillations and energy surges when dealing with continuous and drastic dynamic processes such as water-land transitions, increasing component wear and even leading to transient failures. Furthermore, existing technologies generally treat environmental perception, fault diagnosis, and adaptive control as independent modules, lacking deep collaboration and failing to form a closed-loop intelligent system capable of learning from experience and continuously optimizing its own structure. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a fault prediction method for the amphibious robot's power system based on the differences between water and land operating conditions. It employs a control strategy that integrates dynamic topology modeling, forward-looking risk assessment, and structural learning adaptation, which can improve the proactive fault prediction capability and adaptive stability during the water-land environment switching process, thereby enhancing the robot's overall reliability and resilience.

[0006] The above objectives can be achieved through the following approach: A fault prediction method for amphibious robot propulsion systems based on differences in water and land conditions includes real-time perception of the physical field characteristics of water and land, which is then applied to the energy transmission network and dynamic constraints of the propulsion system. By altering the energy flow distribution pattern and the effective interaction between units, an environmental coupling topology emerges, coupling the current environment. Based on this environmental coupling topology, the energy and information exchange flow between the current water and land physical fields is continuously monitored. The degree of deviation from conservation laws or intrinsic symmetry is quantified, and the flow pattern divergence of the inherent self-consistency and stability margin of the coupling state is obtained. The nonlinear accelerated growth of the flow pattern divergence is a precursor to faults. The method receives navigation intentions to identify impending water-land environment transitions, which act as a symmetry breaking mechanism on the environmental coupling topology, inducing... The nonlinear relaxation of internal energy and information flow spontaneously tends towards a new convergent steady state that maximally suppresses the growth of the flow pattern divergence. Using the new convergent steady state as the initial condition, the instantaneous values ​​of the real-time monitored flow pattern divergence and global energy dissipation rate are used as the evolutionary driving potential. The internal energy and information flow distribution is continuously adjusted to minimize the evolutionary driving potential, guiding the environmental coupled topology state to dynamically generate a minimum action trajectory. The process of traversing the water-land switching process along the minimum action trajectory is used as an effective survival imprint, driving the adjustment of the internal dynamic parameters or connection weights of the environmental coupled topology state. This reduces the evolutionary barrier that needs to be overcome when traversing the switching interface along similar trajectories in the future, achieving structural adaptation and resilient internalization to the differences between water and land conditions.

[0007] Optionally, the emergence of the environmental coupling topology state coupled with the current environment includes: real-time sensing of the physical field characteristics of the water and land and converting them into a non-uniform external field potential of the energy transmission network of the dynamic system; using the external field potential to change the energy conduction characteristics of each transmission path and the interaction between units to form a field-induced modulation parameter set; based on the field-induced modulation parameter set, adaptive adjustment is performed by exchanging energy and information flow, spontaneously tending towards local resonance and synchronization, generating a unit resonance spectrum; based on the field-induced modulation parameter set and the unit resonance spectrum, the dynamic structure of the current environmental physical constraints emerges at the macroscopic level, obtaining the environmental coupling topology state.

[0008] Optionally, obtaining the flow pattern divergence that provides the inherent self-consistency and stability margin of the coupled state includes: deriving the local conservation law and inherent symmetry constraint of energy and information exchange flow at the water-land physical field interface based on the environmental coupled topological state, and quantifying the degree of deviation to generate a local unbalanced flow field; calculating the information entropy generation rate of the local unbalanced flow field in the spatiotemporal dimension to obtain the symmetry-broken entropy flow; and performing a nonlinear coupling operation between the intensity norm of the local unbalanced flow field and the symmetry-broken entropy flow to obtain the flow pattern divergence, wherein the magnitude and growth rate of the flow pattern divergence characterize the degree to which the coupled state deviates from the self-consistent steady state and the rate at which it tends towards instability.

[0009] Optionally, the symmetry-breaking entropy flow includes: local symmetry-breaking intensity and non-equilibrium entropy generation rate, wherein: the local symmetry-breaking intensity is used to quantify the instantaneous deviation of the local unbalanced flow field from the local conservation law or intrinsic symmetry constraint at each node within the water-land physical field interface and the dynamic system; the non-equilibrium entropy generation rate is used to characterize the growth rate of the overall disorder driven by the local symmetry-breaking intensity over time.

[0010] Optionally, the spontaneous tendency to maximally suppress the growth of the divergence of the flow pattern to a new convergent steady state includes: transforming the water-land environment switching information into symmetry breaking on the environmental coupling topology, changing the local stability conditions of internal energy and information flow, and generating a transient instability potential diagram; responding to the transient instability potential diagram, performing nonlinear recombination and relaxation according to the minimum impedance or maximum entropy generation, exploring the relaxation flow pattern that dissipates the symmetry breaking the fastest; and using the relaxation flow pattern to converge to a new dynamic equilibrium point to generate a new convergent steady state.

[0011] Optionally, the method further includes: performing spatiotemporal superposition and nonlinear interferometry analysis on the local unbalanced flow field and the transient instability potential diagram to quantify the coupling amplification effect of the current unbalanced state and the expected switching disturbance, and generating a coupled disturbance interferogram; analyzing the singular structure and gradient region in the coupled disturbance interferogram, and calculating the critical risk index of the water-land switching process in combination with the current growth rate of the flow pattern divergence.

[0012] Optionally, guiding the dynamic generation of the minimum action trajectory of the coupled environmental topology includes: establishing the evolutionary driving potential as a global optimization constraint, driving the coupled environmental topology to evolve naturally in the direction of decreasing instantaneous divergence or dissipation rate, forming an intrinsic evolutionary tendency; constructing a damping field that adjusts the ease of transfer between different regions in the coupled environmental topology based on the switching critical risk index, wherein the damping field exhibits high damping characteristics in regions with high risk indices; and adjusting the effective conductivity or permeability of the internal energy and information flow transmission paths so that the internal driving flux represented by the intrinsic evolutionary tendency is redirected under the action of the damping field, spontaneously relaxing along the path of minimum damping in the damping field, thereby generating a minimum action trajectory.

[0013] Optionally, the formation of intrinsic evolutionary tendency includes: transforming the evolutionary driving potential into a non-uniform potential field on the environmental coupled topological state space, forming a high potential energy region in areas with high instantaneous divergence or dissipation rate; and using the environmental coupled topological state in the non-uniform potential field to spontaneously generate an internal driving flux pointing in the direction of the fastest potential energy reduction, thereby generating an intrinsic evolutionary tendency.

[0014] Optionally, the realization of structural adaptation and resilient internalization to differences in water and land conditions includes: using the minimum action trajectory as an intrinsic template, identifying activated internal dynamic parameters or connection weights in the environmental coupling topology, and generating a set of reinforcement markers; based on the set of reinforcement markers, driving an asymmetric structural plastic process, enhancing the corresponding energy and information transmission efficiency, and relatively suppressing irrelevant paths to form a differentiated transmission topology; based on the differentiated transmission topology, when encountering similar differences in water and land conditions in the future, the internal energy and information flow spontaneously tends to follow the reinforced path, achieving structural adaptation and resilient internalization.

[0015] Based on the same inventive concept, this invention also provides a fault prediction system for an amphibious robot power system based on differences in water and land conditions. The system includes: an environmental coupling topology state emergence module, used to perceive the physical field characteristics of water and land in real time and act on the energy transmission network and dynamic constraints of the power system. By changing the energy flow distribution pattern and the effective interaction between units, an environmental coupling topology state coupling the current environment emerges; a flow pattern divergence monitoring module, used to continuously monitor the energy and information exchange flow between the current water and land physical fields based on the environmental coupling topology state, quantify the degree of deviation from conservation laws or intrinsic symmetry, and obtain the flow pattern divergence of the inherent self-consistency and stability margin of the coupling state, wherein the nonlinear accelerated growth of the flow pattern divergence is a precursor to fault; and a new steady-state spontaneous convergence module, used to receive navigation intentions and identify the impending water-land environment switch as a symmetry breaking event. The system acts on the coupled topological state of the environment and induces nonlinear relaxation of internal energy and information flow, spontaneously tending towards a new convergent steady state that maximally suppresses the growth of the flow pattern divergence. A minimum action trajectory generation module, using the new convergent steady state as an initial condition, takes the instantaneous values ​​of the real-time monitored flow pattern divergence and global energy dissipation rate as the evolutionary driving potential, continuously adjusting the internal energy and information flow distribution to minimize the evolutionary driving potential, guiding the coupled topological state of the environment to dynamically generate a minimum action trajectory. A structural adaptation and resilience internalization module, using the experience of traversing the water-land switching process along the minimum action trajectory as an effective survival imprint, drives the adjustment of the internal dynamic parameters or connection weights of the coupled topological state of the environment, reducing the evolutionary barrier that needs to be overcome when traversing the switching interface along similar trajectories in the future, thus achieving structural adaptation and resilience internalization to the differences in water-land conditions.

[0016] Compared with the prior art, the present invention has the following advantages: 1. By constructing a dynamic environmental coupled topology state and introducing flow pattern divergence as a fault precursor indicator, real-time and accurate assessment is achieved. Compared with traditional methods that rely on static models or single thresholds, this method can capture subtle early anomalies from the overall energy and information flow level, improving the sensitivity and accuracy of fault prediction, providing effective early warning before faults occur, and enhancing the operational safety of the robot. 2. Through a forward-looking adaptive adjustment and dynamic path optimization mechanism, the adaptability and stability of the amphibious robot during the transition between water and land environments are improved. By actively pre-stressing and avoiding risks during the spontaneous convergence of the new steady state and the generation of the minimum action trajectory, the robot can complete the transition of working conditions in a smooth and low-consumption manner, effectively avoiding energy shocks and component overloads caused by sudden environmental changes, and ensuring reliable operation in critical stages; 3. By introducing structural adaptation and resilience internalization mechanisms, the robot system is endowed with a biological-like ability to learn from experience and evolve on its own. The system can solidify successful switching experiences into optimizations of its internal dynamic structure, thereby enabling faster and more robust responses with lower computational costs and energy consumption when facing similar tasks in the future. This improves the robot's long-term autonomy, reliability, and adaptability to specific operating environments.

[0017] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the amphibious robot power system fault prediction method based on the differences between land and water conditions, according to an embodiment of the present invention.

[0020] Figure 2 This is a distribution diagram of the flow pattern divergence at different operating stages according to an embodiment of the present invention.

[0021] Figure 3 This is a schematic diagram showing the association of key technical features in an embodiment of the present invention.

[0022] Figure 4This is a schematic diagram of the structure of the amphibious robot power system fault prediction system based on the difference between land and water conditions according to an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0024] Reference Figure 1 One embodiment of the present invention proposes a fault prediction method for the amphibious robot power system based on the difference between water and land conditions. It adopts a control strategy that integrates dynamic topology modeling, forward-looking risk assessment and structural learning adaptation, which can improve the active fault prediction capability and adaptive stability during the water-land environment switching process, and enhance the overall reliability and resilience of the robot.

[0025] The method described in this embodiment specifically includes: S1. Real-time perception of the physical field characteristics of water and land, and its application to the energy transmission network and dynamic constraints of the power system. By changing the energy flow distribution pattern and the effective interaction between units, an environmental coupling topology state emerges that is coupled with the current environment. Optionally, the emerging environmental coupling topology state coupled with the current environment includes: Real-time sensing of the physical field characteristics of water and land is transformed into a non-uniform external field potential of the power system energy transmission network. The external field potential is used to change the energy conduction characteristics of each transmission path and the interaction between units to form a set of field-induced modulation parameters. Based on the field-induced modulation parameter set, adaptive adjustment is performed by exchanging energy and information flow, spontaneously tending towards local resonance and synchronization, and generating a unit resonance spectrum; Based on the field-induced modulation parameter set and the unit resonance spectrum, the dynamic structure of the current environmental physical constraints emerges at the macroscopic level, resulting in the environmental coupled topological state.

[0026] Specifically, the aim is to construct a dynamic functional network model that reflects the external environment. A multimodal sensor array deployed on the robot body, such as water pressure sensors, sonar, or lidar, senses the physical field characteristics of water or land in real time, acquiring raw data such as water flow velocity, ground slope, or soil friction coefficient. The processing device transforms this characteristic data into a non-uniform external potential through nonlinear mapping. Conceptually, this external potential assigns an abstract potential energy value to each functional unit, characterizing the obstacles or benefits of the environment. This nonlinear mapping is established through offline experiments and simulations. The processing device uses this external potential to adjust the energy conduction characteristics of each transmission path in the energy transmission network and the interaction strength between units, forming a field-induced modulation parameter set. Based on this parameter set, a unit resonance spectrum is generated through adaptive adjustment. Under environmental modulation, energy and information flows are reconstructed, and each functional unit spontaneously tends towards local resonance and synchronous operation. The processing device continuously monitors the state variables of each unit and calculates the synchronization strength index between them, such as using phase-locking values. When synchronization occurs between units, they are considered to have formed a strong functional resonance relationship. All these strong resonance relationships collectively constitute a graph-structured data model on a macroscopic level, namely, a unit resonance spectrum. Based on the field-induced modulation parameter set and the unit resonance spectrum, an environmental coupling topology emerges. The processing device nonlinearly fuses the parameter set and the spectrum. For example, the final effective connection strength between two units may be a weighted combination of their physical connection parameters and synchronization strength. The weighting coefficients here can be determined based on prior knowledge, for example, 0.4 and 0.6. Through this fusion, a dynamic, weighted, directed network graph is ultimately generated, namely, the environmental coupling topology, which characterizes the real and effective energy and information flow pattern under specific environmental constraints.

[0027] For example, suppose an amphibious robot is moving from a calm, static body of water, environment A, to a 15-degree sandy beach, environment B. In environment A, sensors detect a water resistance of 10 Newtons. At this time, the non-uniform external potential generated by the processing device through a mapping model is low, the field-induced modulation parameters are concentrated, and the "equivalent load parameter" of the four wheel-side drive motors M1 to M4 is 0.1. When the robot enters environment B, sensors detect a ground slope of 15 degrees and a sand friction coefficient of 0.5. The processing device immediately transforms these high resistance characteristics into a high-potential external potential through nonlinear mapping. This causes the field-induced modulation parameters to concentrate, and the "equivalent load parameter" of motors M1 to M4 increases from 0.1 to 0.7. Based on this new parameter set, to overcome the high load, the internal controllers of the four drive motors begin to adaptively adjust, and their drive current frequencies spontaneously tend to be consistent. The processing device monitors that the phase lock value between each pair of motors M1 to M4 rapidly increases from 0.3 in environment A to 0.9. In the unit resonance spectrum, the processing device established strong connections between nodes M1, M2, M3, and M4, indicating that they form a tight "land climbing driven cluster." Finally, the processing device fused the parameter set and the spectrum. Using weighting coefficients of 0.4 and 0.6, the new effective connection weight between M1 and M2 was calculated as: (0.7 × 0.4) + (0.9 × 0.6) = 0.28 + 0.54 = 0.82. This high-weight topology represents the environmental coupling topology state that emerges when the robot adapts to land climbing conditions.

[0028] S2. Based on the environmental coupling topology, continuously monitor the energy and information exchange flow between the current water and land physical fields, quantify the degree of deviation from the conservation law or intrinsic symmetry, and obtain the flow pattern divergence of the intrinsic self-consistency and stability margin of the coupling state. The nonlinear accelerated growth of the flow pattern divergence is a precursor to the fault. Optionally, the flow mode divergence used to obtain the intrinsic self-consistency and stability margin of the coupled state includes: Based on the aforementioned environmental coupled topological state, the local conservation law and intrinsic symmetry constraint of energy and information exchange flow at the water-land physical field interface are derived, and the degree of deviation is quantified to generate a local unbalanced flow field. Calculate the information entropy generation rate of the local unbalanced flow field in the spatiotemporal dimension to obtain the symmetric broken entropy flow. The intensity norm of the local unbalanced flow field is nonlinearly coupled with the symmetric broken entropy flow to obtain the flow pattern divergence. The magnitude and growth rate of the flow pattern divergence characterize the degree to which the coupled state deviates from the self-consistent steady state and the rate at which it tends to become unstable.

[0029] Specifically, the aim is to quantify complex operational states into a key indicator characterizing their health and stability margin: flow pattern divergence. Based on the coupled topology of the environment, this diagnostic feature is ultimately generated. First, a locally imbalanced flow field is generated. Based on this dynamic network model of the coupled topology, an ideal baseline for energy and information exchange flows is established. This baseline is composed of local conservation laws and inherent symmetry constraints. The processing device calculates a theoretical healthy flux value for each node and connecting edge in the coupled topology based on these constraints. Simultaneously, actual flux values ​​are obtained using real operational data collected by sensors. By subtracting the theoretical flux value from the actual flux value point by point, the processing device quantifies the deviation of each local location from the ideal state. All these deviations together constitute a vector field, i.e., the locally imbalanced flow field. Based on the obtained locally imbalanced flow field, the symmetry-breaking entropy flow is obtained. The locally imbalanced flow field only represents static deviations; its dynamic evolution trend is key to predicting failures. This step captures this dynamic trend by calculating the information entropy generation rate of the locally imbalanced flow field in the spatiotemporal dimension. When the flow field becomes unstable, it diffuses or fluctuates violently, leading to an accelerated increase in overall disorder. The processing device analyzes the time-series changes and spatial gradients of the localized unbalanced flow field to calculate the growth rate of this disorder, which is the symmetric breaking entropy flow. Finally, the flow pattern divergence is obtained. The information obtained in the first two steps is nonlinearly fused. The processing device extracts the intensity norm and the symmetric breaking entropy flow of the localized unbalanced flow field. To comprehensively reflect the risk of amplitude and velocity, the processing device uses a nonlinear coupling operation to combine the two, for example, multiplying the intensity norm by the exponential term of the symmetric breaking entropy flow. This operation includes an optimization coefficient, obtained through regression training on a historical fault database, to ensure that the flow pattern divergence has the highest sensitivity to early, minor faults. This exponential coupling ensures that the flow pattern divergence value shows a significant increase in both small-amplitude but high-velocity and slow-velocity but large-amplitude anomalies. This nonlinear accelerated growth becomes a clear and reliable precursor signal of a fault, such as... Figure 2 As shown, this figure uses a violin plot combined with box plots and jitter scatter plots to demonstrate the statistical characteristics of the flow pattern divergence: the value is low and concentrated in the "stable operation" stage, the value increases and the distribution becomes more divergent in the "pre-fault" stage, and the value converges to a high level in the "fault state" stage.

[0030] Optionally, the symmetry-breaking entropy flow includes: local symmetry-breaking intensity and non-equilibrium entropy generation rate, wherein: The local symmetry breaking intensity is used to quantify the instantaneous deviation of the local unbalanced flow field from the local conservation law or intrinsic symmetry constraint at each node of the water-land physical field interface and within the dynamic system. The non-equilibrium entropy generation rate is used to characterize the rate of increase of the overall disorder over time driven by the intensity of local symmetry breaking.

[0031] Specifically, this implementation decomposes the symmetry-breaking entropy flow into two core physical quantities for analysis: the static local symmetry-breaking intensity and the dynamic non-equilibrium entropy generation rate. The first step is to quantify the local symmetry-breaking intensity. This step aims to transform the vectorized local imbalance flow field generated in the previous step into a scalar value to characterize the overall magnitude of the dynamic system's deviation from the ideal state at the current instant. The processing device, based on the environmental coupled topology, traverses key nodes within the dynamic system, particularly interaction nodes at the land-water physical field interface and internally high-load functional units. For each node, the processing device extracts its corresponding loss vector in the local imbalance flow field and calculates its magnitude or norm. For example, this is done by calculating the Euclidean distance between the imbalance vector and the ideal zero defined by the local conservation law or intrinsic symmetry constraints. The processing device aggregates the instantaneous deviation magnitudes calculated at all these nodes, for example, using weighted summation or root mean square methods, ultimately obtaining a single scalar value, which is the local symmetry-breaking intensity. The next step is to calculate the non-equilibrium entropy generation rate. This step characterizes the growth rate of the overall disorder of the dynamical system over time, driven by the intensity of local symmetry breaking, which itself constitutes the core of the symmetry-breaking entropy flow. The processing device is not satisfied with a static snapshot of the local symmetry breaking intensity but performs a time-series analysis. It continuously monitors the time-series changes of the local symmetry breaking intensity and calculates its growth rate, for example, by calculating its first-order time derivative or the slope of change within a short time window. A positively increasing local symmetry breaking intensity means that the imbalance within the dynamical system has not been effectively dissipated or suppressed, but is instead accumulating or spreading. This irreversible disorder growth process driven by internal imbalance is quantified as the rate of non-equilibrium entropy generation. In this way, the processing device explicitly correlates a static deviation magnitude with a dynamic deterioration trend, thus providing crucial dynamic input for subsequent computation of flow pattern divergence.

[0032] For example, a robot climbs a 15-degree sandy slope, its environmental coupled topology being a land-climbing driven cluster. At this point, the inherent symmetry constraint requires that the theoretical healthy current of the four drive motors M1 to M4 is 50 amps. At time T0, the actual currents collected by the sensors are M1=50A, M2=51A, M3=50A, and M4=49A. The processing device calculates the local unbalanced flow field as [0A, +1A, 0A, -1A]. The processing device then performs quantization of the local symmetry breaking strength, for example, calculating the L1 norm, obtaining the local symmetry breaking strength = |0|+|1|+|0|+|-1|=2.0A. Between T0 and T1 seconds, this value fluctuates smoothly around 2.0A, and the processing device calculates that the non-equilibrium entropy generation rate is close to 0. Assuming the tuning coefficient is 2.0, the flow pattern divergence at time T1 = 2.0 × exp(2.0 × 0) ≈ 2.0. Subsequently, the bearings of motor M2 began to experience early wear due to sand ingress. At time T2, to maintain the rotational speed, the current in M2 increased to 60A, with the actual current being [50A, 60A, 50A, 49A]. The localized unbalanced flow field became [0A, +10A, 0A, -1A]. The localized symmetry breaking strength = |0| + |10| + |0| + |-1| = 11.0A. Wear was accelerating. At time T3, the current in M2 increased to 75A, with the actual current being [50A, 75A, 50A, 49A]. The localized unbalanced flow field became [0A, +25A, 0A, -1A]. The localized symmetry breaking strength = |0| + |25| + |0| + |-1| = 26.0A. The processing device calculates the non-equilibrium entropy generation rate at time T3, which is the growth rate of the local symmetry breaking intensity from T2 to T3, and obtains a normalized rate value, for example, 0.8. Finally, at time T3, the processing device calculates the flow mode divergence: 26.0A × exp(2.0 × 0.8) = 26.0 × exp(1.6) ≈ 26.0 × 4.95 = 128.7. The flow mode divergence accelerates nonlinearly from 2.0 in the healthy state to 128.7, which the processing device determines as a precursor to a fault.

[0033] S3. Receiving navigation intent identifies the impending switch between land and water environments, which acts as a symmetry breaking mechanism on the coupled topological state of the environment, and induces nonlinear relaxation of internal energy and information flow, spontaneously tending towards a new convergent steady state that maximally suppresses the growth of the flow pattern divergence. Optionally, the spontaneous tendency to maximally suppress the growth of the flow pattern divergence includes the following new convergent steady state: The information on the switching between land and water environments is transformed into symmetry breaking on the coupled topological state of the environment, thereby changing the local stability conditions of internal energy and information flow and generating a transient instability potential diagram. In response to the transient instability potential diagram, nonlinear recombination and relaxation are performed following the minimum impedance or maximum entropy generation, exploring the relaxation flow mode that dissipates the symmetry breaking the fastest. The relaxed flow mode is used to converge to a new dynamic equilibrium point, generating a new convergent steady state.

[0034] Specifically, the system aims to proactively guide pre-adaptive adjustments in response to an impending switch between water and land environments, ensuring a smooth transition to a new, stable operating state and preventing malfunctions caused by sudden environmental changes. The processing device receives information about the impending water-land environment switch from the navigation module, such as the robot about to move from the water surface to the shore. This information is transformed into a symmetry breaking of the current coupled topological state. Based on the physical characteristics of the target environment, the processing device calculates a predicted non-uniform external potential and differs it from the current non-uniform external potential. This differencing potential is applied to the current coupled topological state, disrupting its original equilibrium and altering the local stability conditions of internal energy and information flow. By performing a perturbation analysis on the stability of each node in the topological state, the processing device calculates the potential instability caused by this switch and maps these values ​​onto the topological state, forming a scalar field—the transient instability potential map. Based on this transient instability potential map, an exploratory relaxation flow mode is executed. Responding to the high-potential-energy regions presented in the transient instability potential map, the processing device guides the internal energy and information flow through nonlinear recombination and relaxation. This relaxation process follows the physical principle of minimum impedance or maximum entropy. The processing device explores different internal flow recombination schemes through rapid iterative calculations and evaluates the dissipation rate of each scheme on the transient instability potential diagram. Ultimately, it determines the recombination scheme that can reduce the overall instability potential energy most quickly; this scheme is the relaxor flow pattern. Using the explored relaxor flow pattern, a new convergent steady state is generated. The processing device uses this relaxor flow pattern as a set of dynamic control commands and issues them to each execution unit. For example, before switching to land, based on the relaxor flow pattern's instructions, the power of the underwater thrusters is reduced in advance, while the motors of the land-based drive wheels are activated. The actual operating state smoothly deviates from the current steady state and evolves along the trajectory planned by the relaxor flow pattern towards a new dynamic equilibrium point, eventually converging to the new convergent steady state. This steady state has been pre-adapted to the constraints of the new environment and, in the process, suppresses to the greatest extent possible the drastic fluctuations that could lead to nonlinear growth in the flow pattern divergence.

[0035] Optionally, the method further includes: The local unbalanced flow field and the transient instability potential diagram are spatiotemporally superimposed and nonlinearly interferometrically analyzed to quantify the coupling amplification effect between the current unbalanced state and the expected switching disturbance, and to generate a coupled disturbance interferogram. By analyzing the singular structures and gradient regions in the coupled perturbation interferogram and combining them with the current growth rate of the flow pattern divergence, the critical risk index for the land-water switching process is calculated.

[0036] Specifically, the processing equipment acquires the local unbalanced flow field and transient instability potential map. The equipment performs spatiotemporal superposition and nonlinear interferometry analysis on these two fields defined on the coupled topological state of the environment. The core of this analysis is to quantify the coupling amplification effect, that is, at the same node of the topological state, a large current imbalance quantity encounters a large expected disturbance quantity, which may generate a nonlinear joint risk. The processing equipment performs point-by-point coupling operations on these two fields using nonlinear functions, such as tensor product or kernel function operations, to obtain a completely new data field, which is the coupled disturbance interferogram. The processing equipment analyzes this coupled disturbance interferogram to identify singular structures and gradient regions. Simultaneously, the processing equipment introduces the current growth rate of the flow pattern divergence. Through a comprehensive evaluation model, the critical risk index for the water-land switching process is calculated, as follows: , in, The critical risk index for switching is the final output value of the calculation, used to quantify the overall risk of the switching process. This represents the peak value extracted from the coupled perturbation interferogram. This value corresponds to the singular structure in the coupled perturbation interferogram and characterizes the local risk point where the current state is most strongly coupled with the expected perturbation. The current growth rate of the flow pattern divergence is used to characterize the dynamic deterioration trend of the current health status; Representing the weight of the interferogram, it is a dimensionless tuning coefficient used to adjust the contribution of local spatial risk to the total risk index; The growth rate weight is represented by a dimensionless tuning coefficient used to adjust the contribution of dynamic time risk to the total risk index. Both tuning coefficients are obtained through regression training on a historical fault database or through high-fidelity simulation experiments.

[0037] For example, at time T3, the robot's M2 motor bearing wears, resulting in a flow pattern divergence of 128.7, with a current growth rate of 0.8. The locally unbalanced flow field exhibits a significant imbalance of +25A at the M2 motor node. At this point, an intention to switch from land to water is issued, preparing to re-enter the water from the beach. The processing device receives this intention, converts the "land-water" switching information into symmetry breaking, and calculates that the local stability condition of the M2 motor node will drastically change due to the impending high load activation. Therefore, the M2 motor node exhibits an extremely high potential instability value on the transient instability potential diagram. The current locally unbalanced flow field is spatiotemporally superimposed with the transient instability potential diagram. The processing device finds that the current actual fault point completely coincides with the expected maximum disturbance point, generating strong nonlinear interference. Analyzing the diagram yields an extremely high peak value for the coupled disturbance interferogram, for example, 50.0. The processing device substitutes these values ​​into a formula to calculate the switching critical risk index. Assume the interferogram weight is 1.5 and the growth rate weight is 2.0. The critical risk index for switching is calculated as (1.5 × 50.0) + (2.0 × 0.8) = 75.0 + 1.6 = 76.6. This extremely high critical risk index of 76.6 indicates that if a forced switch between land and water is performed, the existing fault in motor M2 will be coupled and amplified with the switching shock, potentially leading to a power system collapse. Due to the excessively high risk index, the processing equipment rejects the immediate switching action and, based on the principle of "maximally suppressing the growth of flow pattern divergence," generates a relaxed flow pattern aimed at isolating the fault. The processing equipment issues a command to cut off the energy supply to motor M2 and guide motors M1, M3, and M4 to asymmetrically increase their power to maintain the robot's basic stability on the beach. This "land safety steady state" where motor M2 is isolated is the new convergent steady state. In this steady state, the flow pattern divergence is suppressed, avoiding the risk.

[0038] S4. Using the new convergent steady state as the initial condition, the instantaneous values ​​of the real-time monitored flow pattern divergence and global energy dissipation rate are used as the evolution driving potential. The internal energy and information flow distribution is continuously adjusted to minimize the evolution driving potential, guiding the environmental coupled topology state to dynamically generate the minimum action trajectory. Optionally, guiding the dynamic generation of the minimum action trajectory of the coupled topological state of the environment includes: The evolutionary driving potential is established as a global optimization constraint, driving the environmental coupled topological state to evolve naturally in the direction of reducing instantaneous divergence or dissipation rate, forming an intrinsic evolutionary tendency. Based on the switching critical risk index, a damping field is constructed to adjust the ease of transfer between different regions in the environmental coupling topology, wherein the damping field exhibits high damping characteristics in regions with high risk index. By adjusting the effective conductivity or permeability of the internal energy and information flow transmission paths, the internal driving flux represented by the intrinsic evolutionary tendency is redirected under the action of the damping field, and spontaneously relaxes along the path of minimum damping in the damping field, generating a minimum action trajectory.

[0039] Specifically, this step is the core of the dynamic optimization of the entire switching process, aiming to guide the power system to evolve in the way with the lowest energy consumption and highest stability throughout the transition dynamic path. First, it executes the formation of an intrinsic evolutionary tendency. This step establishes the evolutionary driving potential as a global optimization constraint. The processing device takes minimizing this evolutionary driving potential as its objective; in the state space, this is equivalent to defining a cost function. The processing device spontaneously generates an adjustment tendency that tends towards the direction that minimizes this cost function the fastest; this is the intrinsic evolutionary tendency. Second, it executes the construction of a damping field. To mitigate risk, the processing device constructs a damping field in the state space of the coupled topological state based on this index. This damping field is a scalar field; in regions with high switching critical risk indices, the damping value is correspondingly high, forming high-risk "barrier zones," making it difficult for the power system state to traverse these regions. Finally, it executes the generation of a minimum action trajectory, combining the intrinsic evolutionary tendency with the damping field. The processing device redirects the internal driving flux, represented by its intrinsic evolutionary tendency, by continuously adjusting the effective conductivity or permeability of the internal energy and information flow transmission paths. When the internal driving flux encounters a highly damped region, it is deflected and spontaneously relaxes along the path that minimizes damping in the damped field. The dynamic system state depicts a continuous evolutionary path in the state space, which is the minimum action trajectory and the optimal dynamic path for traversing the water-land switching process.

[0040] Optionally, the formation of intrinsic evolutionary tendencies includes: The evolutionary driving potential is transformed into a non-uniform potential field on the environmental coupled topological state space, forming a high potential energy region in areas with high instantaneous divergence or dissipation rate. By utilizing the environmental coupled topological state in the non-uniform potential field, an internal driving flux pointing towards the direction of the fastest potential energy reduction is spontaneously generated, thus generating an intrinsic evolutionary tendency.

[0041] Specifically, by constructing a virtual non-uniform potential field, the abstract optimization objective is visualized as a physical force that can guide state evolution, providing a driving force for the generation of the minimum action trajectory. First, the evolutionary driving potential is transformed into a non-uniform potential field. The real-time monitored evolutionary driving potential is used as input. This potential, representing "cost" or "risk," is mapped onto the environmental coupled topology state space of all possible states. The mapping rule is that, in the state space, for dynamic system state points corresponding to high flow pattern divergence or high global energy dissipation rate, their potential energy value in the non-uniform potential field is also assigned a higher value, thus forming a high potential energy region; conversely, stable and energy-efficient state points correspond to low potential energy regions. Second, the internal driving flux is spontaneously generated using the non-uniform potential field. According to physical principles, dynamic systems in a non-uniform potential field will spontaneously tend to move in the direction of decreasing potential energy. The environmental coupled topology state in this non-uniform potential field will also spontaneously generate an internal driving flux pointing in the direction of the fastest decrease in potential energy. Mathematically, this internal driving flux is equal to the negative gradient of the non-uniform potential field. At each point in the state space, the processing device calculates the potential field gradient at that point and takes its opposite direction to obtain a vector. This vector represents the direction and intensity of the dynamic system's adjustment to reduce its "cost" in that state. This internal driving flux, pointing in the direction of the fastest reduction in potential energy, is directly defined as the intrinsic evolutionary tendency of the final output, providing clear guidance for subsequent dynamic adjustments.

[0042] For example, the robot is in a new convergent steady state of a "land-climbing driven cluster" and receives a switching command to return to the water. The goal of the processing device is to generate an optimal minimum action trajectory to complete this "land-water" switch. First, the processing device transforms the evolutionary driving potential into a non-uniform potential field. In this potential field, the "land" state is in a high potential energy region due to the high motor load, while the "water" state is in a low potential energy region due to the propeller cruising. This creates an intrinsic evolutionary tendency toward the "water" state, i.e., a desire to switch as quickly as possible to save energy. Simultaneously, the processing device calculates the switching critical risk index. Calculations revealed that a "hard switch" path within 0.5 seconds would traverse a high-risk zone. Therefore, the processing equipment constructed a damped field with high damping characteristics within this high-risk zone. Finally, the intrinsic evolutionary tendency, driven by energy conservation, was redirected by this damped field when attempting a "hard switch." The flux was forced to relax along a path that minimized damping, thus bypassing the "hard switch" path. The specific operation corresponding to this generated minimum action trajectory is: smoothly reducing the power of the four drive motors while simultaneously smoothly increasing the power of the underwater thruster within 5 seconds. This "power overlap" path is the optimal solution for traversing the switching process.

[0043] S5. Following the minimum action trajectory through the water-land switching process, and using it as an effective survival imprint, drive the adjustment of the internal dynamic parameters or connection weights of the environmental coupled topology state, reduce the evolutionary barrier that needs to be overcome when following similar trajectories through the switching interface in the future, and realize the structural adaptation and resilient internalization of the differences between water and land conditions.

[0044] Optionally, the achievement of structural adaptation and resilient internalization to differences in water and land conditions includes: Using the minimum action trajectory as an intrinsic template, the activated internal dynamic parameters or connection weights in the environmental coupled topology are identified to generate a reinforcement tag set. Based on the aforementioned enhanced marker set, an asymmetric structural plasticity process is driven, thereby enhancing the corresponding energy information conduction efficiency and relatively suppressing irrelevant paths, thus forming a differentiated conduction topology. Based on the aforementioned differentiated transmission topology, when encountering similar differences in water and land conditions in the future, the internal energy and information flow will spontaneously tend to follow the reinforced path, achieving structural adaptation and resilient internalization.

[0045] Specifically, the optimal switching experience is solidified into the "instinct" of the dynamic system, achieving a sublimation from "computation" to "adaptation." First, a reinforcement mark set is generated. When the dynamic system successfully completes a land-water switch along the minimum action trajectory, this trajectory is considered an intrinsic template and a valid survival imprint. The processing device back-analyzes this trajectory to identify which internal dynamic parameters or connection weights in the environmental coupled topology were most frequently activated or contributed the most to the trajectory during the process of achieving this optimal path. For example, certain specific energy transmission paths or control loops. These identified key parameters and weights are aggregated to form the reinforcement mark set. Second, based on the reinforcement mark set, an asymmetric structural plasticity process is executed. The processing device initiates an adjustment mechanism to modify the basis parameters of the environmental coupled topology. For parameters and paths marked by the reinforcement mark set, the processing device enhances their energy information transmission efficiency, for example, by increasing their connection weights or decreasing their transmission damping. Irrelevant paths that were not used or contributed little in the optimal trajectory may be relatively suppressed. This asymmetric adjustment process transforms the environmental coupling topology from a general network into a network "specialized" for specific switching scenarios, thus forming a differentiated transmission topology. Finally, based on this differentiated transmission topology, structural adaptation and resilient internalization are achieved. Through this plasticity process, the basic structure undergoes permanent or semi-permanent changes. In the future, when the robot encounters similar differences in water and land conditions, its internal energy and information flow, evolving within the new differentiated transmission topology, will spontaneously tend to follow the reinforced path due to the "preference" of its physical structure. This is equivalent to transforming a complex, computationally demanding "optimal path" into a structural "highway," thereby reducing the evolutionary barriers that need to be overcome to reproduce this optimal switch in the future, achieving structural adaptation and resilient internalization to differences in water and land conditions.

[0046] For example, the robot successfully executed a minimum action trajectory with a 5-second "power overlap," completing a smooth "land-to-water" switch. The processing device uses this 5-second "power overlap" trajectory as a valid survival imprint. The processing device analyzes this trajectory and generates a reinforced marker set containing dynamic parameters controlling the smooth decrease in power of the four drive motors and the smooth increase in power of the underwater thrusters, particularly the connection weights used to coordinate the overlap timing. Subsequently, based on this reinforced marker set, the processing device drives an asymmetric structural plasticity process. The processing device permanently increases this specific set of connection weights by 20%, forming a new differentiated conduction topology. The next time the robot performs a "land-to-water" switch, because this set of connection weights has been reinforced, its internal energy and information flow will naturally and spontaneously tend to reproduce this 5-second "power overlap" path during evolution, as it has become the "minimum damped" path in the entire state space. This eliminates the need for complex global optimization calculations, achieving structural adaptation and resilience internalization of the switch process. Figure 3 As shown, a circular bar chart illustrates the simulated attributes of eleven key technical features across three dimensions: the length of the bar represents computational resource consumption, the gray level of the bar represents the frequency of calls, and the height of the black dots connected by the dashed lines inside represents stability gain.

[0047] Based on the same inventive concept, this invention also provides a fault prediction system for the power system of amphibious robots based on differences in water and land conditions, such as... Figure 4 As shown, the system includes: The environmental coupled topology state emergence module is used to sense the physical field characteristics of water and land in real time and act on the energy transmission network and dynamic constraints of the power system. By changing the energy flow distribution pattern and the effective interaction between units, the environmental coupled topology state coupled with the current environment emerges. The flow pattern divergence monitoring module is used to continuously monitor the energy and information exchange flow between the current water and land physical fields based on the environmental coupling topology state, quantify the degree of deviation from the conservation law or intrinsic symmetry, and obtain the flow pattern divergence of the intrinsic self-consistency and stability margin of the coupling state. The nonlinear accelerated growth of the flow pattern divergence is a precursor to the fault. The new steady-state spontaneous convergence module is used to receive navigation intent recognition of the upcoming water and land environment switch, as a symmetry breaking action on the coupled topological state of the environment, and induce nonlinear relaxation of internal energy and information flow, spontaneously tending to a new convergence steady state that maximally suppresses the growth of the flow pattern divergence. The minimum action trajectory generation module is used to take the new convergent steady state as the initial condition, use the instantaneous values ​​of the real-time monitored flow pattern divergence and global energy dissipation rate as the evolution driving potential, continuously adjust the internal energy and information flow distribution to minimize the evolution driving potential, and guide the environmental coupled topology state to dynamically generate the minimum action trajectory. The structural adaptation and resilience internalization module is used to drive the adjustment of the internal dynamic parameters or connection weights of the environmental coupled topology state by traversing the water-land switching process along the minimum action trajectory as an effective survival imprint. This reduces the evolutionary barrier that needs to be overcome when traversing the switching interface along similar trajectories in the future, thereby realizing structural adaptation and resilience internalization to the differences between water and land conditions.

[0048] To verify the feasibility and effectiveness of the invention, it was applied to a commercially available amphibious reconnaissance robot to solve the problem of power system failure during complex water-land transitions.

[0049] In one test, while the robot was cruising in water, the processing equipment sensed the water flow and the slope of the land ahead through sensors, generating a set of field-induced modulation parameters and generating an environmental coupled topological state centered on the thrusters. One thruster simulated encountering entanglement with aquatic plants, disrupting the symmetry of the dynamic system. The processing equipment immediately generated a locally unbalanced flow field based on the deviation of the local conservation law and calculated the continuous increase of the symmetry-broken entropy flow. The nonlinear coupling of these two factors led to a nonlinear acceleration of the flow pattern divergence, triggering a fault precursor warning. The robot received the navigation intention of "landing on the mudflats." The processing equipment converted this intention into a transient instability potential diagram. The processing equipment performed a nonlinear interferometric analysis on this transient instability potential diagram and the existing locally unbalanced flow field, calculating an extremely high switching critical risk index. To maximally suppress the growth of the flow pattern divergence, the processing equipment induced it to spontaneously tend towards a new convergent steady state, i.e., isolating the faulty thruster and operating safely in an asymmetric dynamic mode, rather than forcibly landing. In another landing under healthy conditions, the evolutionary driving potential formed an intrinsic evolutionary tendency for rapid switching. However, the damping field redirected the dynamic system, causing it to follow a path of minimum damping and generate a smooth minimum action trajectory, i.e., a "power smooth overlap" switching mode. This successful "power smooth overlap" trajectory was used as an effective survival imprint. Based on this, the processing equipment generated a set of reinforcement markers, driving an asymmetric structural plastic process and forming a differentiated conduction topology. In subsequent tasks, the internal energy flow spontaneously tended towards this reinforced path, achieving structural adaptation and resilience internalization for this operating condition.

[0050] It should be noted that the functional division and information interaction between the various modules described above are logical, but in terms of physical implementation, they can be integrated on the same software platform or deployed in a distributed manner. The connections between them represent data flow and control flow, aiming to collaboratively achieve the objectives of this invention. The above descriptions are merely exemplary embodiments of this invention and should not be construed as limiting the scope of protection of this invention.

Claims

1. A method for predicting the faults of an amphibious robot's power system based on the differences between land and water conditions, characterized in that, The method includes: Real-time perception of the physical field characteristics of water and land, and its application to the energy transmission network and dynamic constraints of the power system, by changing the energy flow distribution pattern and the effective interaction between units, gives rise to an environmental coupling topology state that is coupled with the current environment. Based on the environmental coupling topology, the energy and information exchange flow between the current land and water physical fields is continuously monitored, the degree of deviation from the conservation law or intrinsic symmetry is quantified, and the flow pattern divergence of the inherent self-consistency and stability margin of the coupling state is obtained. The nonlinear accelerated growth of the flow pattern divergence is a precursor to the fault. The system receives navigation intent and identifies the impending switch between land and water environments. This is then used as a symmetry-breaking mechanism to act on the coupled topological state of the environment, and induces nonlinear relaxation of internal energy and information flow, spontaneously tending towards a new convergent steady state that maximally suppresses the growth of the flow pattern divergence. Using the new convergent steady state as the initial condition, the instantaneous values ​​of the real-time monitored flow pattern divergence and global energy dissipation rate are used as the evolutionary driving potential. The internal energy and information flow distribution is continuously adjusted to minimize the evolutionary driving potential, guiding the environmental coupled topology state to dynamically generate the minimum action trajectory. The process of traversing the water-land transition along the minimum action trajectory will serve as an effective survival imprint, driving the adjustment of the internal dynamic parameters or connection weights of the environmental coupled topology state. This will reduce the evolutionary barrier that needs to be overcome when traversing the transition interface along similar trajectories in the future, thereby achieving structural adaptation and resilient internalization to the differences between water and land conditions.

2. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 1, characterized in that, The emerging environmental coupling topology states that couple the current environment include: Real-time sensing of the physical field characteristics of water and land is transformed into a non-uniform external field potential of the power system energy transmission network. The external field potential is used to change the energy conduction characteristics of each transmission path and the interaction between units to form a set of field-induced modulation parameters. Based on the field-induced modulation parameter set, adaptive adjustment is performed by exchanging energy and information flow, spontaneously tending towards local resonance and synchronization, and generating a unit resonance spectrum; Based on the field-induced modulation parameter set and the unit resonance spectrum, the dynamic structure of the current environmental physical constraints emerges at the macroscopic level, resulting in the environmental coupled topological state.

3. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 1, characterized in that, The flow mode divergence used to obtain the intrinsic self-consistency and stability margin of the coupled state includes: Based on the aforementioned environmental coupled topological state, the local conservation law and intrinsic symmetry constraint of energy and information exchange flow at the water-land physical field interface are derived, and the degree of deviation is quantified to generate a local unbalanced flow field. Calculate the information entropy generation rate of the local unbalanced flow field in the spatiotemporal dimension to obtain the symmetric broken entropy flow. The intensity norm of the local unbalanced flow field is nonlinearly coupled with the symmetric broken entropy flow to obtain the flow pattern divergence. The magnitude and growth rate of the flow pattern divergence characterize the degree to which the coupled state deviates from the self-consistent steady state and the rate at which it tends to become unstable.

4. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 3, characterized in that, The symmetry-breaking entropy flow includes: local symmetry-breaking intensity and non-equilibrium entropy generation rate, wherein: The local symmetry breaking intensity is used to quantify the instantaneous deviation of the local unbalanced flow field from the local conservation law or intrinsic symmetry constraint at each node of the water-land physical field interface and within the dynamic system. The non-equilibrium entropy generation rate is used to characterize the rate of increase of the overall disorder over time driven by the intensity of local symmetry breaking.

5. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 3, characterized in that, The new convergent steady state that spontaneously tends to maximally suppress the growth of the flow pattern divergence includes: The information on the switching between land and water environments is transformed into symmetry breaking on the coupled topological state of the environment, thereby changing the local stability conditions of internal energy and information flow and generating a transient instability potential diagram. In response to the transient instability potential diagram, nonlinear recombination and relaxation are performed following the minimum impedance or maximum entropy generation, exploring the relaxation flow mode that dissipates the symmetry breaking the fastest. The relaxed flow mode is used to converge to a new dynamic equilibrium point, generating a new convergent steady state.

6. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 5, characterized in that, The method further includes: The local unbalanced flow field and the transient instability potential diagram are spatiotemporally superimposed and nonlinearly interferometrically analyzed to quantify the coupling amplification effect between the current unbalanced state and the expected switching disturbance, and to generate a coupled disturbance interferogram. By analyzing the singular structures and gradient regions in the coupled perturbation interferogram and combining them with the current growth rate of the flow pattern divergence, the critical risk index for the land-water switching process is calculated.

7. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 6, characterized in that, The process of guiding the dynamic generation of the minimum action trajectory in the coupled topological state of the environment includes: The evolutionary driving potential is established as a global optimization constraint, driving the environmental coupled topological state to evolve naturally in the direction of reducing instantaneous divergence or dissipation rate, forming an intrinsic evolutionary tendency. Based on the switching critical risk index, a damping field is constructed to adjust the ease of transfer between different regions in the environmental coupling topology, wherein the damping field exhibits high damping characteristics in regions with high risk index. By adjusting the effective conductivity or permeability of the internal energy and information flow transmission paths, the internal driving flux represented by the intrinsic evolutionary tendency is redirected under the action of the damping field, and spontaneously relaxes along the path of minimum damping in the damping field, generating a minimum action trajectory.

8. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions as described in claim 7, characterized in that, The formation of intrinsic evolutionary tendencies includes: The evolutionary driving potential is transformed into a non-uniform potential field on the environmental coupled topological state space, forming a high potential energy region in areas with high instantaneous divergence or dissipation rate. By utilizing the environmental coupled topological state in the non-uniform potential field, an internal driving flux pointing towards the direction of the fastest potential energy reduction is spontaneously generated, thus generating an intrinsic evolutionary tendency.

9. The method for predicting the fault of an amphibious robot's power system based on the difference between land and water conditions according to claim 1, characterized in that, The structural adaptation and resilient internalization to the differences between land and water conditions include: Using the minimum action trajectory as an intrinsic template, the activated internal dynamic parameters or connection weights in the environmental coupled topology are identified to generate a reinforcement tag set. Based on the aforementioned enhanced marker set, an asymmetric structural plasticity process is driven, thereby enhancing the corresponding energy information conduction efficiency and relatively suppressing irrelevant paths, thus forming a differentiated conduction topology. Based on the aforementioned differentiated transmission topology, when encountering similar differences in water and land conditions in the future, the internal energy and information flow will spontaneously tend to follow the reinforced path, achieving structural adaptation and resilient internalization.

10. A fault prediction system for an amphibious robot power system based on differences in water and land conditions, applied to the fault prediction method for an amphibious robot power system based on differences in water and land conditions as described in any one of claims 1-9, characterized in that, The system includes: The environmental coupled topology state emergence module is used to sense the physical field characteristics of water and land in real time and act on the energy transmission network and dynamic constraints of the power system. By changing the energy flow distribution pattern and the effective interaction between units, the environmental coupled topology state coupled with the current environment emerges. The flow pattern divergence monitoring module is used to continuously monitor the energy and information exchange flow between the current water and land physical fields based on the environmental coupling topology state, quantify the degree of deviation from the conservation law or intrinsic symmetry, and obtain the flow pattern divergence of the intrinsic self-consistency and stability margin of the coupling state. The nonlinear accelerated growth of the flow pattern divergence is a precursor to the fault. The new steady-state spontaneous convergence module is used to receive navigation intent recognition of the upcoming water and land environment switch, as a symmetry breaking action on the coupled topological state of the environment, and induce nonlinear relaxation of internal energy and information flow, spontaneously tending to a new convergence steady state that maximally suppresses the growth of the flow pattern divergence. The minimum action trajectory generation module is used to take the new convergent steady state as the initial condition, use the instantaneous values ​​of the real-time monitored flow pattern divergence and global energy dissipation rate as the evolution driving potential, continuously adjust the internal energy and information flow distribution to minimize the evolution driving potential, and guide the environmental coupled topology state to dynamically generate the minimum action trajectory. The structural adaptation and resilience internalization module is used to drive the adjustment of the internal dynamic parameters or connection weights of the environmental coupled topology state by traversing the water-land switching process along the minimum action trajectory as an effective survival imprint. This reduces the evolutionary barrier that needs to be overcome when traversing the switching interface along similar trajectories in the future, thereby realizing structural adaptation and resilience internalization to the differences between water and land conditions.

Images