Cable laying clamping fault early warning method and system based on multi-pressure sensing fusion

Abstract

The invention discloses a cable laying clamping fault early warning method and system based on multi-pressure sensing fusion. The method comprises the following steps: acquiring data of a plurality of pressure sensors of a plurality of cable conveyors and corresponding cable laying clamping fault data; the assignment module carries out weight assignment on the fault sensitivity degree according to the data correlation of the sensor; inputting the weighted pressure sensor data into an encoder for fusion to obtain a fault indication state quantity; and inputting the fault indication state quantities at different moments into the LSTM for network parameter training, performing loss function calculation according to the fault prediction output of the LSTM and the fault data, and if a loss function value does not meet a condition, performing retraining until the loss function value meets the condition, so as to obtain the fault indication state quantity of the LSTM. And the current assignment module, the encoder and the LSTM are used as a cable laying clamping fault early warning model to carry out fault early warning. According to the invention, the clamping fault in the cable laying process can be effectively and automatically warned in advance.

Description

Technical Field

[0001] The present invention belongs to the technical field of cable laying safety, and relates to a cable laying clamping fault early warning method and system based on multi-pressure sensor fusion. Background Art

[0002] During power cable construction, key quality control nodes are the cable reel payout, conveying, and turning sections, including payout racks, cable conveyors, traction machines, force measurement units, and intelligent master control units. The cable conveyor is crucial equipment for cable laying operations, clamping the cable on tracks for transport. Multiple cable conveyors may be required during cable laying, and each is equipped with pressure sensors to prevent laying failures caused by problems with traction, guidance, and clamping.

[0003] Current cable conveying and pulling equipment has a low level of intelligence. Clamping during cable conveyance requires manual operation. Excessive clamping force can cause frictional damage to the cable, while insufficient clamping force can lead to slippage and squeezing during conveyance. Construction quality depends on the operator's skill. The pressure sensing thresholds used by multiple cable conveyors are unclear, making it difficult to effectively provide early warning of clamping failures during cable laying. Therefore, it is necessary to propose a solution that integrates multiple pressure sensors to provide automatic early warning of cable laying clamping failures. Summary of the Invention

[0004] In order to solve the deficiencies in the prior art, the present invention provides a cable laying clamping fault early warning method and system based on multi-pressure sensing fusion.

[0005] The present invention adopts the following technical solutions.

[0006] The cable laying clamping fault early warning method based on multi-pressure sensor fusion includes:

[0007] S1. Acquire data from multiple pressure sensors of multiple cable conveyors and corresponding cable laying and clamping fault data;

[0008] S2, an assignment module assigns weights of fault sensitivity according to the correlation of multiple pressure sensor data of multiple cable conveyors to obtain weighted pressure sensor data;

[0009] S3, inputting the weighted pressure sensor data into the encoder for fusion, and obtaining a fault indication state quantity after the pressure sensor data is fused;

[0010] S4. Input the fault indication state quantities at different times into LSTM to train network parameters, and calculate the loss function based on the fault prediction output of LSTM and the cable laying and clamping fault data;

[0011] S5. If the loss function value does not meet the conditions, return to S4 until the loss function value meets the conditions. The assignment module, encoder and LSTM at this time are used as the cable laying and clamping fault warning model to perform fault warning.

[0012] Preferably, in the assignment module, the weighted pressure sensor data is:

[0013] F i ′(t)=c i *F i (t), i=1,…,n

[0014]

[0015] Among them, c i The weight assigned to the i-th pressure sensor;

[0016] C i 、C j is the sum of the correlation coefficients of the i-th and j-th pressure sensor data and other sensor data;

[0017] F i (t), F j (t), F k (t) is the data of the i, j, and k pressure sensors at time t,

[0018] is the data mean of the i-th, j-th, and k-th pressure sensors;

[0019] N is the total number of pressure sensors; n is the total data sampling time.

[0020] Preferably, the encoder comprises an input layer, a hidden layer and an output layer, and the input layer, the hidden layer and the output layer are fully connected neural networks.

[0021] Preferably, the encoder inputs the weighted pressure sensor data into the encoder for fusion, and the formula for obtaining the fault indication state quantity after the pressure sensor data is fused is:

[0022]

[0023] Among them, F t ′ is the weighted pressure sensor data vector at time t;

[0024] Z t is the fault indication state quantity at time t;

[0025] H t is the output of the hidden layer node at time t;

[0026] W h 、b h is the hidden layer gating weight and bias matrix;

[0027] W z 、b z is the output layer gating weight and bias matrix;

[0028] σ is the sigmoid function.

[0029] Preferably, the LSTM includes a forget gate, an input gate, and an output gate, and introduces a time step attention mechanism to reflect the degree of influence of different time steps on output prediction; wherein the time step attention mechanism is:

[0030]

[0031] where γ t is the weight factor; T represents transposition; β t is the attention weight;

[0032] V β , W β , U β , W d and b d is the fully connected bias parameter;

[0033] h t is the hidden state vector at time t; r t The final prediction output of LSTM.

[0034] Preferably, the hidden state vector at time t is specifically:

[0035]

[0036] Where: f t 、i t 、o t is the output of the forget gate, input gate, and output gate at time t;

[0037] h t and G t are the hidden state vector and unit state vector at time t respectively;

[0038] h t-1 and G t-1 are the hidden state vector and unit state vector at time t-1 respectively;

[0039] W f 、W i 、W o 、W C They are the trainable forget gate, input gate, output gate, and weight matrix of the LSTM unit;

[0040] b f 、b i 、b o 、b C They are the trainable forget gate, input gate, output gate, and bias matrix of the LSTM unit;

[0041] Z t is the fault indication state quantity at time t; tanh(·) is the hyperbolic tangent function; σ is the sigmoid function.

[0042] Preferably, the loss function is:

[0043]

[0044] Where: ω is the loss judgment parameter, R t is the cable laying clamping fault data at time t, r t is the fault prediction output value of LSTM at time t.

[0045] Preferably, the condition is: the loss function value is less than 10 -4 .

[0046] The cable laying clamping fault early warning system based on multi-pressure sensor fusion includes:

[0047] A data acquisition module, used to acquire data from multiple pressure sensors of multiple cable conveyors and corresponding cable laying and clamping fault data;

[0048] An assignment module, configured to assign weights of fault sensitivity according to correlation of data from multiple pressure sensors of multiple cable conveyors, and obtain weighted pressure sensor data;

[0049] An encoder is used to fuse the weighted pressure sensor data to obtain a fault indication state quantity after the pressure sensor data is fused;

[0050] The network training module is used to input the fault indication state quantity at different times into the LSTM for network parameter training, and calculate the loss function based on the fault prediction output of the LSTM and the cable laying and clamping fault data;

[0051] The fault warning module is used to return to the network training module to retrain parameters when the loss function value does not meet the conditions until the loss function value meets the conditions. At this time, the assignment module, encoder and LSTM are used as the cable laying and clamping fault warning model to perform fault warning.

[0052] A terminal includes a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the method.

[0053] A computer-readable storage medium stores a computer program, which implements the steps of the method when executed by a processor.

[0054] The beneficial effects of the present invention are as follows:

[0055] The assignment module of the present invention assigns weights according to the proportion of the sum of the correlation coefficients of each pressure sensor data and other sensor data in the total correlation coefficient. By allocating coefficients according to the degree of correlation of each sensor in the whole, the sensor that can best reflect the installation status can be given the largest weight, which helps to improve the accuracy of early warning. The encoder of the present invention receives data after weight allocation, which can better reflect the status after fusion.

[0056] The present invention introduces an attention mechanism into LSTM to reflect the influence of different time steps on output prediction, making the prediction results more accurate. It also adopts an improved loss function that takes robustness into consideration, reduces the penalty for abnormal points, increases robustness and convergence stability, and can effectively provide automatic early warning of clamping faults in the cable laying process. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flowchart of the cable laying and clamping fault early warning method based on multi-pressure sensor fusion of the present invention;

[0058] Figure 2 This is a flow chart of the cable laying clamping fault early warning method based on multi-pressure sensor fusion of the present invention;

[0059] Figure 3 Schematic diagram of the encoder structure of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0061] like Figure 1-Figure 2 As shown, embodiment 1 of the present invention provides a cable laying clamping fault early warning method based on multi-pressure sensor fusion, comprising:

[0062] S1. Acquire data from multiple pressure sensors of multiple cable conveyors and corresponding cable laying and clamping fault data;

[0063] Further preferably, a database is constructed by using the data of multiple pressure sensors of multiple cable conveyors that change over time and the corresponding relationship between the data and the faults;

[0064] First, the time-varying data of multiple pressure sensors of multiple cable conveyors (F1(t), F2(t), F3(t)...F n (t)) and fault (R t )Build a database based on the corresponding relationship.

[0065] Taking a 110kV AC cable direct buried installation as an example, a database is first constructed based on the data of the four pressure sensors of two cable conveyors that change with the installation time and the corresponding relationship with the faults; the data of the pressure sensors that change with time are F1(t), F2(t), F3(t), F4(t), and the fault data R t . Understandably, R t The value ranges from 0 to 1, and the probability of failure can be calculated linearly according to the time distance of the impending failure, which is R t value.

[0066] S2, an assignment module assigns weights of fault sensitivity according to the correlation of multiple pressure sensor data of multiple cable conveyors to obtain weighted pressure sensor data;

[0067] Furthermore, preferably, the sensitivity of different conveyor pressure sensors to faults is reflected by assigning weights to the pressure sensor data of different conveyors in the database. Specifically analyzing the correlation between each pressure sensor data and the data of other sensors, and assigning weights based on the proportion of the sum of the correlation coefficients of each pressure sensor data with the other sensor data in the total correlation coefficient, has the advantage of assigning coefficients based on the overall correlation of each sensor, giving the highest weight to the sensor that best reflects the installation status, which helps improve the accuracy of early warnings.

[0068] The weight assignment formula is:

[0069]

[0070] Among them, c i The weight assigned to the i-th pressure sensor; C i is the sum of the correlation coefficients of the i-th pressure sensor data and other sensor data; C j is the sum of the correlation coefficients of the jth pressure sensor data and other sensor data; is the sum of the correlation coefficients of all sensors with other sensor data;

[0071] F i (t) is the data of the i-th pressure sensor at time t, is the mean value of the i-th pressure sensor data; F j (t) is the data of the j-th pressure sensor at time t, is the mean value of the j-th pressure sensor data; F k (t) is the data of the kth pressure sensor at time t, is the mean value of the kth pressure sensor data;

[0072] N is the total number of pressure sensors; n is the total sampling time; i, j, and k are the serial numbers of pressure sensors.

[0073] The weighted pressure sensor data is:

[0074] F i ′(t)=c i *F i (t)(i=1,…,n)

[0075] The subscript i represents the i-th pressure sensor, and n represents the number of pressure sensors.

[0076] S3, inputting the weighted pressure sensor data into the encoder for fusion, and obtaining a fault indication state quantity after the pressure sensor data is fused;

[0077] The weighted multi-pressure sensor data is input into the encoder, which outputs the fault indication state quantity after the multi-pressure sensor data is fused. The encoder is one of the artificial intelligence networks, and its main function is to compress and process the data. The encoder is constructed by several layers of fully connected neural networks, including hidden layers and output layers. The output layer outputs the fault indication state quantity Z after the multi-pressure sensor data is fused. t .

[0078] In this embodiment, the four weighted pressure sensor data are input into the encoder, which includes an input layer, a hidden layer and an output layer. Figure 3 The input layer, hidden layer and output layer are fully connected neural networks. The number of neurons in the input layer is the same as the number of weighted pressure sensors, which is 4 here. The number of neurons in the hidden layer is set to 2, and the number of neurons in the output layer is 1. The fault indication state Z after the fusion of multiple pressure sensor data is output. t .

[0079] The encoder inputs the weighted pressure sensor data into the encoder for fusion. The formula for obtaining the fault indication state quantity after the pressure sensor data fusion is:

[0080]

[0081] Among them, F t ′ is the weighted pressure sensor data vector at time t, that is, Fi ′(t) set; Z t is the fault indication state quantity at time t; H t is the output of the hidden layer node at time t; W h 、b h is the hidden layer gating weight and bias matrix; W z 、b z is the output layer gating weight and bias matrix; σ is the Sigmoid function.

[0082] S4. Input the fault indication state quantities at different times into LSTM to train network parameters, and calculate the loss function based on the fault prediction output of LSTM and the cable laying and clamping fault data;

[0083] S5. If the loss function value does not meet the conditions, return to S4 until the loss function value meets the conditions. The assignment module, encoder, and LSTM at this time are used as the cable laying and clamping fault warning model to perform fault warning. That is, after the training is completed, the assignment module, encoder, and LSTM form the warning model. When in use, the real-time pressure sensor data of multiple cable conveyors is directly input into the warning model. The model automatically assigns weights, encodes, and predicts, and issues fault warnings based on the output value, where the output value is a value between 0 and 1, indicating the probability of failure. The LSTM model parameters are updated during training, while the assignment module and encoder are not updated. S4-S5 are described in detail as follows:

[0084] (1) In addition to the forget gate, input gate, and output gate, the LSTM of the present invention introduces an attention mechanism to reflect the degree of influence of different time steps on the output prediction, making the prediction result more accurate.

[0085]

[0086] Where: f t 、i t 、o t is the output of the forget gate, input gate, and output gate at time t; h t and G t is the hidden state vector and unit state vector at time t; W f and b f is the trainable forget gate weight and bias matrix; W i and b i is the trainable input gate weight and bias matrix; W o and b o is the trainable output gate weight and bias matrix; W C 、b C is the weight and bias matrix of the trainable LSTM unit; Z t is the fault indication state quantity at time t; ht-1 and G t-1 are the hidden state vector and the cell state vector at time t-1; tanh(·) is the hyperbolic tangent function; and σ is the sigmoid function, which outputs a number between zero and one, describing each component that can pass through the LSTM cell. The sigmoid layer determines which values ​​will be updated, while the tanh layer creates a new candidate value that can be added to the LSTM cell state vector.

[0087] In order to integrate the impact of the data of each monitoring quantity at different time steps on the future prediction time step data, the hidden state vector h of the LSTM network is t and the cell state vector G t The attention mechanism intelligently adjusts time-step weights. By calculating attention and activating the softmax layer, the sum of all weights is equal to 1. This determines the degree to which fault indication parameters at different times influence other times. The attention weights are then combined with the input fault indication parameters through weighted multiplication and fusion, yielding an output that reflects the relationship between the fault indication parameters at different times. Finally, a fully connected layer is used to output a fault warning.

[0088] The time step attention is calculated as follows

[0089]

[0090] where γ t is the weight factor; T represents transposition; β t is the attention weight, V β , W β , U β , W d and b d is the fully connected bias parameter, r t is the final prediction output.

[0091] (2) Output the LSTM fault prediction r t and the fault data R in the database t The loss function is calculated to train the LSTM network parameters, where the loss function is an improved loss function considering robustness:

[0092]

[0093] Where: ω is the loss judgment parameter. When the deviation is less than ω, the square loss function is used. When it is greater than ω, the linear loss function is used. ω is set according to the specific problem and can generally be 0.1 to 10. t is the cable laying clamping fault data at time t (R t The value ranges from 0 to 1, and the probability is calculated linearly according to the time distance of the impending fault. tis the fault prediction output value of LSTM at time t.

[0094] (3) Determine the error, that is, whether the loss function value is within the preset acceptable range: If so, the fault prediction output of LSTM can be used as future prediction, representing the fused key variables reflecting the clamping fault; if not, continue to train the above algorithm using the known database.

[0095] The fault indication state quantities at different moments are input into the long short-term memory deep neural network (LSTM). In this embodiment, the fault indication state quantities at 10 moments are input into the LSTM network to predict the fault prediction output r at the next 5 moments. t , by adding r t With fault data R t Calculate the error (L) and train the LSTM network parameters. When the error is less than a certain limit, which is 10 in this embodiment, -4 , model training is completed.

[0096] The trained early warning model is used to automatically assign weights and encode the fault indication state quantities at different times into LSTM, which can be used according to the output value r t The value from 0 to 1 represents the probability of failure. The warning threshold can be selected according to the actual engineering margin. For example, in this embodiment, 0.8 is selected as the alarm threshold.

[0097] Embodiment 2 of the present invention provides a cable laying and clamping fault early warning system based on multi-pressure sensor fusion, comprising:

[0098] A data acquisition module, used to acquire data from multiple pressure sensors of multiple cable conveyors and corresponding cable laying and clamping fault data;

[0099] An assignment module, configured to assign weights of fault sensitivity according to correlation of data from multiple pressure sensors of multiple cable conveyors, and obtain weighted pressure sensor data;

[0100] An encoder is used to fuse the weighted pressure sensor data to obtain a fault indication state quantity after the pressure sensor data is fused;

[0101] The network training module is used to input the fault indication state quantity at different times into the LSTM for network parameter training, and calculate the loss function based on the fault prediction output of the LSTM and the cable laying and clamping fault data;

[0102] The fault warning module is used to return to the network training module to retrain parameters when the loss function value does not meet the conditions until the loss function value meets the conditions. At this time, the assignment module, encoder and LSTM are used as the cable laying and clamping fault warning model to perform fault warning.

[0103] A terminal includes a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the method.

[0104] A computer-readable storage medium stores a computer program, which implements the steps of the method when executed by a processor.

[0105] The beneficial effects of the present invention are as follows:

[0106] The assignment module of the present invention assigns weights according to the proportion of the sum of the correlation coefficients of each pressure sensor data and other sensor data in the total correlation coefficient. By allocating coefficients according to the degree of correlation of each sensor in the whole, the sensor that can best reflect the installation status can be given the largest weight, which helps to improve the accuracy of early warning. The encoder of the present invention receives data after weight allocation, which can better reflect the status after fusion.

[0107] The present invention introduces an attention mechanism into LSTM to reflect the influence of different time steps on output prediction, making the prediction results more accurate. It also adopts an improved loss function that takes robustness into consideration, reduces the penalty for abnormal points, increases robustness and convergence stability, and can effectively provide automatic early warning of clamping faults in the cable laying process.

[0108] The present disclosure may be a system, method and / or computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0109] A computer-readable storage medium can be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination thereof. As used herein, a computer-readable storage medium is not to be construed as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through an electrical wire.

[0110] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions to be stored in the computer-readable storage medium in each computing / processing device.

[0111] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., utilizing an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions. The electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A cable laying clamping fault early warning method based on multi-pressure sensor fusion, characterized in that: include: S1. Acquire multiple pressure sensor data of multiple cable conveyors and corresponding cable laying and clamping fault data; S2, an assignment module assigns weights of fault sensitivity according to the correlation of multiple pressure sensor data of multiple cable conveyors to obtain weighted pressure sensor data; S3, inputting the weighted pressure sensor data into the encoder for fusion, and obtaining the fault indication state quantity after the pressure sensor data is fused; S4, inputting the fault indication state quantities at different times into LSTM for network parameter training, and calculating the loss function according to the fault prediction output of LSTM and the cable laying and clamping fault data; S5. If the loss function value does not meet the conditions, return to S4 until the loss function value meets the conditions. The assignment module, encoder and LSTM at this time are used as the cable laying and clamping fault warning model to perform fault warning.

2. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 1 is characterized in that: In the assignment module, the weighted pressure sensor data is: F i ′(t)=c i *F i (t),i=1,…,n Among them, c i The weight assigned to the i-th pressure sensor; C i , C j is the sum of the correlation coefficients of the i-th and j-th pressure sensor data and other sensor data; F i (t), F j (t), F k (t) is the data of the i, j, and k pressure sensors at time t, is the data mean of the i-th, j-th, and k-th pressure sensors; N is the total number of pressure sensors; n is the total data sampling time.

3. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 1 is characterized in that: The encoder comprises an input layer, a hidden layer and an output layer, and the input layer, the hidden layer and the output layer are fully connected neural networks.

4. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 1 is characterized in that: The encoder inputs the weighted pressure sensor data into the encoder for fusion, and the formula for obtaining the fault indication state quantity after the pressure sensor data is fused is: Among them, F t ′ is the pressure sensor data vector after weighting at time t; Z t is the fault indication state quantity at time t; H t is the hidden layer node output at time t; W h 、b h is the hidden layer gating weight and bias matrix; W z , b z is the output layer gating weight and bias matrix; σ is the Sigmoid function.

5. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 1 is characterized in that: The LSTM includes a forget gate, an input gate, and an output gate, and introduces a time step attention mechanism to reflect the influence of different time steps on output prediction; the time step attention mechanism is: where γ t is the weight factor; T represents transposition; β t is the attention weight; V β , W β , U β , W d and b d is the fully connected bias parameter; h t is the hidden state vector at time t; r t The final prediction output of LSTM.

6. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 5 is characterized in that: The hidden state vector at time t is specifically: Where: f t 、i t , o t is the output of the forget gate, input gate, and output gate at time t; G t is the unit state vector at time t; h t-1 and G t-1 are the hidden state vector and unit state vector at time t-1 respectively; W f , W i , W o , W C They are the trainable forget gate, input gate, output gate, and weight matrix of the LSTM unit; b f 、b i 、b o 、b C They are the trainable forget gate, input gate, output gate, and bias matrix of the LSTM unit; Z t is the fault indication state quantity at time t; tanh(·) is the hyperbolic tangent function; σ is the Sigmoid function.

7. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 1 is characterized in that: The loss function is: Where: ω is the loss judgment parameter, R t is the cable laying clamping fault data at time t, r t is the fault prediction output value of LSTM at time t.

8. The cable laying clamping fault early warning method based on multi-pressure sensor fusion according to claim 1 is characterized in that: The condition is: the loss function value is less than 10 -4 .

9. A cable laying clamping fault early warning system based on multi-pressure sensor fusion, using the method described in any one of claims 1 to 8, characterized in that: The system comprises: A data acquisition module, used to acquire data of multiple pressure sensors of multiple cable conveyors and corresponding cable laying and clamping fault data; An assignment module, used for assigning weights of fault sensitivity according to the correlation of multiple pressure sensor data of multiple cable conveyors to obtain weighted pressure sensor data; An encoder is used to fuse the weighted pressure sensor data to obtain a fault indication state quantity after the pressure sensor data is fused; The network training module is used to input the fault indication state quantities at different times into the LSTM for network parameter training, and to calculate the loss function based on the fault prediction output of the LSTM and the cable laying and clamping fault data; The fault warning module is used to return to the network training module to re-train parameters when the loss function value does not meet the conditions, until the loss function value meets the conditions. At this time, the assignment module, encoder and LSTM are used as the cable laying and clamping fault warning model to perform fault warning.

10. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images