A fault detection method for an agricultural irrigation solenoid valve controller

Abstract

This invention belongs to the field of data processing technology, specifically relating to a fault detection method for an agricultural irrigation solenoid valve controller. The method includes: synchronously acquiring voltage and current sequences during the driving cycle; using dead-time data from the initial driving phase to identify the total loop parameters, including cable resistance and inductance, online; constructing a physical constraint operator based on these parameters and introducing it as a penalty term into the objective function of variational mode decomposition to perform physical model-driven signal decomposition on the current sequence, thereby extracting pure mechanical mode components; reconstructing the phase space of these components to generate a state point cloud and calculating the topological distance between this point cloud and the ideal point cloud; and comparing this distance with a preset threshold to determine whether the solenoid valve has a mechanical fault. This invention effectively overcomes the time-varying effect of the channel and achieves stable and accurate detection of early, weak mechanical faults.

Description

Technical Field

[0001] This invention relates to the field of data processing technology. Specifically, it relates to a fault detection method for an agricultural irrigation solenoid valve controller. Background Technology

[0002] In large-scale agricultural irrigation systems, reliable online fault detection of the numerous solenoid valves distributed across the fields is crucial for ensuring the efficient operation of irrigation automation.

[0003] Existing technologies typically rely on analyzing the current and voltage signals of the solenoid valve drive circuit to detect abnormalities in waveform characteristics (such as rise time and peak current) to determine faults such as valve core jamming and coil short circuits.

[0004] However, in large irrigation areas, controllers and actuators are often connected by hundreds of meters of cable. The cable impedance is significantly affected by ambient temperature, causing the total electrical parameters of the circuit to be time-varying and unknown. This inevitably mixes the channel time-varying effects with the characteristics of the directly acquired current signal amplitude, baseline, etc., rather than a pure mechanical state reflection, seriously interfering with the stability of fault criteria.

[0005] Secondly, to extract the implicit mechanical motion information, existing methods may employ purely mathematical signal decomposition techniques to process current waveforms. However, due to the lack of constraints on underlying physical laws (such as circuit dynamics equations), such unsupervised decomposition is prone to mode aliasing or spurious components with unclear physical meaning at low signal-to-noise ratios, making it difficult to robustly separate weak valve core motion features from a strong electrical drive background.

[0006] Therefore, in agricultural irrigation systems with variable environmental conditions and complex signal transmission channels, there is an urgent need for a fault detection method that can identify and compensate for channel time-varying effects online, accurately extract mechanical motion components from composite signals based on physical principles, and maintain robustness. Summary of the Invention

[0007] To address the technical problem that existing technologies struggle to robustly extract early mechanical fault features under complex time-varying channels, this invention provides solutions in several aspects.

[0008] In a first aspect, the present invention provides a fault detection method for an agricultural irrigation solenoid valve controller, comprising:

[0009] Synchronously acquire the power supply voltage sequence and loop current sequence of the solenoid valve drive circuit during the complete drive cycle;

[0010] Based on the voltage and current data during the dead time period when the valve core has not yet moved in the initial stage of the drive, the total loop resistance, including the resistance of the long-distance cable and the coil resistance, and the total loop inductance, including the distributed inductance of the line and the coil inductance, are identified.

[0011] A physical constraint operator is constructed based on the total loop resistance and total loop inductance, and this physical constraint operator is introduced as a penalty term into the objective function of variational mode decomposition to decompose the current sequence, thereby extracting the pure mechanical mode components characterizing the mechanical motion of the valve core.

[0012] The pure mechanical modal components are reconstructed in phase space to generate a measured state point cloud, and the topological landscape distance between the measured state point cloud and the pre-stored fault-free ideal state point cloud is calculated.

[0013] The distance to the topological landscape is compared with a preset fault determination threshold. If the distance exceeds the fault determination threshold, the solenoid valve is determined to have a mechanical fault.

[0014] Preferably, the identification includes: extracting the discrete voltage and current sequences corresponding to the dead time period; establishing a discretized voltage balance equation for the resistor-inductor series circuit; and fitting the voltage balance equation using a nonlinear least squares optimization algorithm to obtain numerical estimates of the total loop resistance and the total loop inductance.

[0015] Preferably, the physical constraint operator includes: at each sampling time, subtracting the sum of the ohmic voltage drop caused by the total loop resistance and the induced voltage drop caused by the total loop inductance from the sampled instantaneous voltage to obtain the residual value.

[0016] Preferably, the objective function includes: the objective function is based on the standard variational mode decomposition objective function, with the addition of a penalty term for the physical constraint operator. This penalty term is used to force the sum of each mode component to satisfy the physical relationship of the resistor-inductor series circuit when solving the objective function.

[0017] Preferably, solving the objective function includes: performing iterative optimization using the alternating direction multiplier method, and initializing the center frequency of the modal components based on prior physical knowledge during the iteration process, until the preset convergence condition is met, and then outputting multiple modal components including the pure mechanical modal component.

[0018] Preferably, the phase space reconstruction includes: calculating the mutual information function of the pure mechanical modal components to determine the optimal delay time; setting the embedding dimension to a fixed value; and reconstructing the one-dimensional mechanical modal time series into a multi-dimensional state space point cloud based on the delay time and the embedding dimension.

[0019] Preferably, the topological landscape distance calculation includes: performing continuous cohomology analysis on the measured state point cloud reconstructed based on the pure mechanical modal components and the pre-stored fault-free ideal state point cloud to obtain the persistence information of their respective one-dimensional ring structures; constructing a persistent landscape function based on the persistence information; and calculating the norm distance between the two persistent landscape functions as the topological landscape distance.

[0020] Preferably, the synchronous acquisition includes: using a synchronous sampling circuit to acquire voltage and current data, with a sampling frequency of not less than 10kHz.

[0021] The beneficial effects of this invention are as follows:

[0022] By utilizing the purely resistive electrical phase before valve core movement to identify channel parameters in real time, interference from changes in the line environment on the signal is effectively eliminated. By incorporating physical laws as constraints into the signal decomposition process, components representing mechanical motion can be accurately separated from complex signals. Combined with a topological feature analysis method that is insensitive to signal scaling and translation, a highly robust and accurate diagnosis of mechanical faults is ultimately achieved. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a fault detection method for an agricultural irrigation solenoid valve controller according to an embodiment of the present invention. Detailed Implementation

[0024] 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, not all, of the embodiments of the present invention. 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.

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0026] This invention discloses a fault detection method for an agricultural irrigation solenoid valve controller, referring to... Figure 1 This includes steps S1-S4:

[0027] S1: Collect the instantaneous voltage sequence at the power supply terminal and the instantaneous current sequence in the circuit during the drive cycle of the solenoid valve, and simultaneously record the ambient temperature data.

[0028] Acquiring the raw signal containing complete electrical and mechanical dynamic information of the solenoid valve is crucial. In agricultural irrigation systems, the operation of the solenoid valve generates weak nonlinear disturbances in the current waveform of the drive circuit due to changes in its internal mechanical states, such as valve core displacement, spring compression, and fluid resistance. To capture these transient characteristics from the electrical signal, it is necessary to acquire the electrical signal.

[0029] The solenoid valve drive circuit is monitored in real time using a high-precision signal acquisition module. This monitoring occurs during the drive cycle when the controller issues an opening command. Internally, an analog-to-digital converter (ADC) is used to synchronously acquire the instantaneous voltage at the power supply terminal. and the instantaneous current of the circuit Sampling frequency Take the experience value That is, sampling interval This is to ensure that high-frequency information of the circuit's electrical signals can be recorded.

[0030] Meanwhile, since the resistivity of long-distance transmission cables is highly sensitive to temperature, it is necessary to use a temperature sensor to synchronously record the current ambient temperature. .

[0031] Due to cable resistance The relationship with temperature follows a physical formula. ,in Temperature coefficient of resistance (TCR) is a physical parameter determined by the material properties. When long-distance transmission cables are made of copper, the TCR is... ,pass A temperature compensation mechanism can be established to prevent impedance drift caused by diurnal temperature variations from being misdiagnosed as device failure. Among other things, For reference temperature, The reference resistance values ​​for the reference temperature are all known parameters determined by the cable material and specifications. An empirical value of 25 degrees Celsius can be used. For a 500-meter-long copper core cable with a cross-sectional area of ​​1.5 square millimeters, its reference resistance at 25 degrees Celsius is... A reasonable empirical value is 5.9 ohms.

[0032] Furthermore, obtain the contents Time series data at each sampling point: voltage sequence and current sequence and scalar ambient temperature .

[0033] S2: Based on voltage and current data during the dead time period when the valve core has not yet moved in the initial stage of the drive, identify the total loop electrical parameters, including the impedance of long-distance cables.

[0034] In large-scale agricultural irrigation scenarios, the cable, sometimes hundreds of meters long, between the controller and the solenoid valve constitutes the main transmission channel, and its impedance... Affected by day and night and seasonal ambient temperature The impact is significant. If it is not possible to obtain information in real time... The total loop electrical parameters, including those of the circuit, cannot effectively distinguish between the affine transformations such as overall scaling and offset caused by changes in line impedance in the current waveform and the nonlinear back electromotive force component generated by the mechanical motion of the valve core, leading to false alarms or missed alarms in subsequent diagnosis.

[0035] To address the uncertainty in the overall loop electrical parameters, during the extremely short time following the application of the drive signal, before the valve core begins to move due to mechanical inertia, the solenoid valve coil is equivalent to a purely resistive coil. and inductor The static load constituted.

[0036] This period of time Internally, no back electromotive force is generated in the circuit due to the mechanical movement of the valve core, and the total current is... The change is entirely due to the power supply voltage Cable resistance and the resistance of the coil and the inductance of the coil Therefore, by analyzing the data during the static load phase, the total loop resistance can be obtained. Total loop inductance ,in Distribute inductance to the line.

[0037] Specifically, acquiring the collected data and Then, it is necessary to extract The corresponding data segment for the time period The value is determined based on the sampling frequency. Calculate the corresponding number of discrete points Number of discrete points The corresponding time length is The value of is determined to ensure complete coverage of the static load of the purely electrical response before the valve core actuates.

[0038] Based on interception For the corresponding data segment, using the backward Euler method, the discrete voltage balance equation can be obtained: ,in The sampling interval is... It is a time period Corresponding to Extracting data segments from the middle, These are the sampling point indices of the discrete sequence, corresponding to the element positions of the difference vector. Calculate the current difference vector. , representing the discrete difference of current with respect to time, and thus the first The current difference vector of each element is .

[0039] At this point, the parameter identification problem is transformed into solving an overdetermined system of linear equations. .

[0040] A numerical optimization algorithm is used to fit the discrete voltage balance equations in order to find the optimal solution. and The Levenberg-Marquardt algorithm was selected for nonlinear least squares optimization.

[0041] Adjust parameters iteratively Minimize the objective function This ensures that after iterative convergence, the output is the total loop resistance accurately identified under the current operating conditions. Total loop inductance The number of discrete points corresponding to the dead time. , This indicates rounding down to the nearest integer.

[0042] S3: Based on the identified total loop electrical parameters, construct dynamic constraints and perform physical model-driven signal decomposition on the acquired current sequence to separate the pure modal components characterizing the mechanical motion of the valve core.

[0043] To accurately extract the weak components characterizing the mechanical motion of the valve core under strong electrical conditions, if the acquired current sequence is directly analyzed... Traditional signal decomposition methods, such as empirical mode decomposition or variational mode decomposition, suffer from mode aliasing problems. Because... The electrical fundamental response driven by the power supply voltage and the nonlinear back electromotive force (Back-EMF) interference signal generated by the valve core movement are nonlinearly superimposed.

[0044] This results in extremely weak mechanical motion components under low signal-to-noise ratio conditions, especially in cables hundreds of meters long with significant thermal noise. Their spectrum is severely aliased with that of the electrical trend components, leading to mode aliasing or the generation of spurious components, making it impossible to extract key fault features.

[0045] Since the dynamic changes of the solenoid valve system follow the laws of circuit dynamics, this means that at any given moment, the power supply voltage... It must be equal to the voltage drop across the resistor. Inductor voltage drop and mechanical back electromotive force The sum of. Among them, It is generated by the mechanical movement of the valve core that serves as a bridge connecting electrical measurements and mechanical status.

[0046] Thus, a dynamic residual operator is constructed. As a physical constraint operator, it is used to calculate the instantaneous deviation of the measured current signal relative to an ideal RL series model of pure electrical circuits without mechanical motion. By introducing the energy of this residual operator as a penalty term into the objective function of signal decomposition, the decomposition algorithm can be forced to search within the feasible solution space that satisfies the basic physical laws of the circuit. This enables physical model-driven blind source separation of electrical driving trends and mechanical motion components, improving the accuracy of extracting spectral information corresponding to key fault features.

[0047] Obtain the total loop resistance obtained in step S2 Total loop inductance , construct the first Dynamic residual operator at each sampling point For the first The operator is defined as follows: First, calculate the induced voltage caused by the current change at each sampling point. Next, calculate the ohmic voltage drop across the resistor. Then the dynamic residual operator Characterized the actual power supply voltage The deviation between the voltage predicted by the pure electrical model and the voltage predicted by the pure electrical model, i.e. .

[0048] In the ideal pure electrical static load phase It should be close to zero; when the valve core begins to move and generates back electromotive force. At that time, according to Kirchhoff's voltage law in physics, we have Therefore, the dynamic residual operator It can characterize the relationship between microscopic changes in electric current and hidden mechanical motion.

[0049] Standard VMD (Variational Mode Decomposition) will input signal Decomposed into A mode component with limited bandwidth Its standard objective function is to minimize the sum of the estimated bandwidths of each modal component, denoted as . .

[0050] To incorporate the energy of the residual operator as a penalty term into the objective function of signal decomposition, forcing the decomposition algorithm to search within the feasible solution space that satisfies the basic physical laws of the circuit, this achieves physical model-driven blind source separation of electrical drive trends and mechanical motion components, thus constructing an enhanced objective function. :

[0051]

[0052] in, The discretized spectral constraint term for standard variational mode decomposition (VMD) is expressed mathematically as follows: In the formula, In theory, the first Discrete-time series of modal components; Indicates the first The center frequency of each modal component; This represents the discrete impulse function (Kronecker delta). This represents the discrete convolution operator. This represents the discrete gradient operator (corresponding to the partial derivative in the continuous field). To avoid confusion with the symbol for current, the symbol used here is... Represents the imaginary unit ( ); This represents the total number of modal components, i.e., the preset number of decomposition layers, which is taken as an empirical value of 5.

[0053] For the dynamic penalty term, where This indicates that the modal superposition signal obtained in the current iteration will be used. As electric current Substituting the dynamic residual operator The residual value calculated at time, where, For the first The modal component in the ... Discrete values ​​of each sampling point.

[0054] This is a hyperparameter used to balance the relative importance of spectral constraints and physical constraints. Its empirical value is set based on the signal-to-noise ratio, and is taken as an empirical value of [value missing]. By introducing a physical penalty term, VMD is forced to ensure that the combination of these modes satisfies the condition as much as possible while searching for narrowband modes. and The fundamental circuit equations are defined.

[0055] To solve the above constrained optimization problem, the Alternating Direction Method of Multipliers (ADMM) is used for iterative calculation.

[0056] During initialization, in addition to randomly initializing each mode... and its center frequency In addition, the initial range of the center frequency is guided based on physical priors.

[0057] For example, the center frequency corresponding to the electric drive trend It can be initialized to near DC 0Hz, while the frequency corresponding to the mechanical motion component may be... Based on the typical mechanical response time constant of the solenoid valve reciprocal To set initial values, experience value .

[0058] In each iteration, the algorithm alternately updates each modality. Center frequency And Lagrange multipliers, to achieve updates Introducing a physical penalty term in the steps The gradient.

[0059] The iterative process stops when the convergence condition is met. For example, all modes in two adjacent iterations... The total change is below the threshold The iteration can stop once either the maximum number of iterations is reached or the maximum number of iterations is satisfied. The maximum number of iterations is taken as an empirical value of 500.

[0060] After stopping iteration when the convergence condition is met, output Modal components By utilizing the temporal morphology, spectral characteristics, and their effect on the physical residuals of each component... The contribution of the valve core can be used to obtain the pure mechanical modal components that characterize the mechanical motion of the valve core. .

[0061] Typically, this component appears synchronously with the rising edge of the drive signal in the time domain, its energy is mainly concentrated in the tens of hertz frequency band, and it has little effect on the physical residual. Its contribution is dominant.

[0062] At the same time, the algorithm will also output a major electrical trend component. Its waveform is smooth and conforms to the exponential increase law of a first-order RL circuit.

[0063] At this point, the original composite current signal Successfully decomposed into noise Electrical trend components and key mechanical motion characteristics .

[0064] S4: Based on the pure modal components, the degree of deviation from the ideal mechanical response in the state space geometry is quantified through topological data analysis, and fault diagnosis is performed accordingly.

[0065] Pure mechanical modal components Although electrical trends are filtered out, its amplitude is still affected by cable impedance. The scaling effect is significant, and the waveform baseline may drift due to environmental electromagnetic interference.

[0066] Simply relying on one-dimensional Euclidean space metrics such as amplitude thresholds or shape similarity makes it difficult to distinguish between impedance changes caused by temperature and dynamic characteristic distortions caused by valve core jamming. Temperature-induced impedance changes are global changes, while dynamic characteristic distortions caused by valve core jamming are layout changes. To improve the robustness of signal feature extraction, dynamic changes are represented by mining the geometric and topological invariants of high-dimensional point clouds.

[0067] Pure mechanical mode components of a one-dimensional time series The system is mapped to a point cloud in a high-dimensional state space using phase space reconstruction technology. The shape of this point cloud reflects the evolution trajectory of the system in phase space. The normal solenoid valve's engagement process should present a smooth, closed loop trajectory in phase space; however, when the valve core experiences mechanical failures such as jamming or obstruction, the topology of this trajectory will be torn, twisted, or develop abnormal holes.

[0068] By calculating the distance between the measured point cloud and a standard point cloud generated by an ideal physical model in terms of topological features, the difference between the two can be represented, thereby distinguishing between impedance changes caused by temperature and dynamic characteristic distortions caused by valve core jamming.

[0069] Furthermore, for the pure mechanical modal components Phase space reconstruction is performed. Reconstruction requires two key parameters: delay time. and embedding dimension To determine the optimal delay time ,calculate Mutual information functions under different lags The optimal delay time is defined as the time at which the mutual information function first reaches a local minimum. This ensures that the reconstructed coordinates remain independent while maintaining a physical and dynamic correlation; for the embedding dimension Because the mechanical motion of the solenoid valve core is mainly governed by inertia, spring force, and electromagnetic force, its independent state variables are limited, thus... Take the experience value as fixed. To construct a three-dimensional phase space.

[0070] Determine parameters and Then, phase space reconstruction is performed to generate a set of measured state point clouds. Indexing at any given time The reconstructed phase vector is This allows us to iterate through the indices of all time series to obtain the point cloud. .

[0071] Simultaneously, an ideal mechanical response sequence is generated for comparison. This sequence can be used to obtain a fault-free, parameter-standard solenoid valve dynamic model through numerical simulation. With the same and Reconstruct the point cloud to obtain the ideal state set. .

[0072] It represents the standard topological structure of the point cloud that the valve core motion should have in the state space under ideal fault-free conditions; and since it is the topological feature of the periodic or quasi-periodic mechanical motion trajectory, it corresponds to the ring structure existing in the point cloud, and thus it is necessary to extract the ring structure feature.

[0073] Obtaining point clouds and Next, to calculate the topological feature differences between the two point clouds, Vietoris-Rips complex filtering is performed on each point cloud: a scale parameter is set. When the Euclidean distance between any two points in the point cloud is less than When they are in a certain state, they are connected by edges to establish a connection; as... from Gradually increase to a maximum value , The empirical value can be taken as the point cloud diameter. The method of constructing the topological structure of the point cloud at different scales through Vietoris-Rips complex filtering to extract persistent information of ring features is a well-known detail and will not be elaborated further.

[0074] Simultaneously, the time from the initial appearance of each ring structure is recorded as the birth scale. The corresponding death scale upon disappearance The scale range it spans, and the duration of its duration This reflects the stability of each ring structure. Based on the duration of all ring structures, a multi-layer scale-domain feature function is constructed. This is called the landscape function.

[0075] The landscape function is specifically constructed as follows: sort all ring structures according to their life cycle length, and select the one with the longest life cycle. One pattern, Take the experience value For each ring structure, a contribution function is constructed within its surviving scale range. Finally, the contribution functions of each layer are superimposed to form the overall landscape function. This maps the complex ring-shaped structural features in point clouds to a scale parameter. The real value on.

[0076] Finally, by comparing the landscape functions of the measured point cloud and the ideal point cloud, the topological difference is calculated as a fault diagnosis index, and the measured landscape function is calculated. With ideal landscape function The difference between them yields the topological landscape distance. : This distance value The larger the value, the more severe the deviation between the measured trajectory and the ideal trajectory in terms of overall ring structure characteristics.

[0077] Preset a fault determination threshold Fault diagnosis is performed, among which Calculated from historical normal data mean Plus Double standard deviation , experience value .like If the condition is met, the solenoid valve is determined to have a mechanical fault; otherwise, the solenoid valve is determined not to have a mechanical fault.

[0078] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A fault detection method for an agricultural irrigation solenoid valve controller, characterized in that, include: Synchronously acquire the power supply voltage sequence and loop current sequence of the solenoid valve drive circuit during the complete drive cycle; Based on the voltage and current data during the dead time period when the valve core has not yet moved in the initial stage of the drive, the total loop resistance, including the resistance of the long-distance cable and the coil resistance, and the total loop inductance, including the distributed inductance of the line and the coil inductance, are identified. A physical constraint operator is constructed based on the total loop resistance and total loop inductance, and this physical constraint operator is introduced as a penalty term into the objective function of variational mode decomposition to decompose the current sequence, thereby extracting the pure mechanical mode components characterizing the mechanical motion of the valve core. The pure mechanical modal components are reconstructed in phase space to generate a measured state point cloud. The topological landscape distance between this measured state point cloud and a pre-stored fault-free ideal state point cloud is calculated. The calculation of the topological landscape distance includes: The measured state point cloud reconstructed based on the pure mechanical modal components is subjected to continuous coherence analysis with the pre-stored fault-free ideal state point cloud to obtain the persistent information of their respective one-dimensional ring structures. Construct a persistent landscape function based on the aforementioned persistent information; Calculate the norm distance between two persistent landscape functions, which is taken as the topological landscape distance; The distance to the topological landscape is compared with a preset fault determination threshold. If the distance exceeds the fault determination threshold, the solenoid valve is determined to have a mechanical fault.

2. The fault detection method for an agricultural irrigation solenoid valve controller according to claim 1, characterized in that, The identification includes: Extract the discrete voltage and current sequences corresponding to the dead time period; Establish the voltage balance equation for a discretized series resistor-inductor circuit; The voltage balance equation is fitted using a nonlinear least squares optimization algorithm to obtain numerical estimates of the total loop resistance and the total loop inductance.

3. The fault detection method for an agricultural irrigation solenoid valve controller according to claim 1, characterized in that, The physical constraint operator includes: at each sampling time, subtracting the sum of the ohmic voltage drop caused by the total loop resistance and the induced voltage drop caused by the total loop inductance from the sampled instantaneous voltage to obtain the residual value.

4. The fault detection method for an agricultural irrigation solenoid valve controller according to claim 2, characterized in that, The objective function includes: The objective function is based on the standard variational mode decomposition objective function, with the addition of a penalty term related to the physical constraint operator. This penalty term is used to force the sum of each mode component to satisfy the physical relationship of the resistor-inductor series circuit when solving the objective function.

5. The fault detection method for an agricultural irrigation solenoid valve controller according to claim 4, characterized in that, Solving the objective function includes: The alternating direction multiplier method is used for iterative optimization. During the iteration process, the center frequency of the modal components is initialized and guided based on prior physical knowledge until the preset convergence condition is met. Then, multiple modal components, including the pure mechanical modal component, are output.

6. The fault detection method for an agricultural irrigation solenoid valve controller according to claim 1, characterized in that, The phase space reconstruction includes: Calculate the mutual information function of the pure mechanical modal components to determine the optimal delay time; Set the embedding dimension to a fixed value; Based on the aforementioned delay time and embedding dimension, the one-dimensional mechanical modal time series is reconstructed into a multi-dimensional state space point cloud.

7. The fault detection method for an agricultural irrigation solenoid valve controller according to claim 1, characterized in that, The synchronous acquisition includes: using a synchronous sampling circuit to acquire voltage and current data, with a sampling frequency of not less than 10kHz.

Images