Teller machine fault diagnosis method and device, equipment, storage medium and product

Abstract

The invention discloses a teller machine fault diagnosis method and device, equipment, a storage medium and a product, and the method comprises the steps: obtaining to-be-diagnosed data, and processing the to-be-diagnosed data; and taking the processed to-be-diagnosed data as input data of the target prototype network, and inputting the to-be-diagnosed data into the trained target prototype network to obtain a diagnosis result output by the target prototype network. According to the teller machine fault diagnosis method disclosed by the invention, the fault diagnosis precision of the teller machine can be improved, so that the possibility of bank outlet business loss caused by the fault of the teller machine is reduced.

Description

Technical Field

[0001] The present invention relates to the field of machine learning technology, and in particular to a method, device, equipment, storage medium and product for diagnosing ATM faults. Background Art

[0002] Automated Teller Machine (ATM) is an important machine for banks to conduct offline business. The core component is the cash dispenser module, which is mainly responsible for cash deposit and withdrawal related services. When it fails, it will directly affect the service quality and customer satisfaction of the branch.

[0003] In the prior art, the vibration signals generated by the rolling bearings during the cash dispensing process are generally collected and processed, and then signal analysis, machine learning, and deep learning methods are used to achieve real-time fault diagnosis of the cash dispensing module of the ATM. However, in actual bank branch scenarios, it often takes a long time for the ATM to transition from a normal operating state to a fault state, and the duration of the fault state is short, making it difficult to collect sufficient fault data, which limits the performance of the fault diagnosis model to a certain extent. Summary of the invention

[0004] The present invention provides a teller machine fault diagnosis method, device, equipment, storage medium and product to achieve high-precision fault diagnosis of the teller machine.

[0005] According to one aspect of the present invention, a method for diagnosing a fault of an ATM is provided, comprising:

[0006] Acquiring data to be diagnosed, and processing the data to be diagnosed;

[0007] The processed data to be diagnosed is used as input data of the target prototype network, and the data to be diagnosed is input into the trained target prototype network to obtain the diagnosis result output by the target prototype network.

[0008] Further, the data to be diagnosed includes a vibration signal of the banknote dispensing module, and processing the data to be diagnosed includes:

[0009] Performing continuous wavelet transform processing on the vibration signal of the banknote dispensing module to obtain a time-frequency diagram of the vibration signal;

[0010] The vibration signal time-frequency diagram is used as the processed data to be diagnosed.

[0011] Furthermore, the target prototype network includes a feature extraction network, a distance measurement network, a fault diagnosis network and a decoder, wherein the feature extraction network acts as an encoder and forms a convolutional autoencoder with the decoder.

[0012] Furthermore, the training method of the target prototype network includes:

[0013] Acquire training data; wherein the training data includes a support data set and a query data set;

[0014] Performing type recognition on the support data set through the target prototype network to obtain at least one category prototype;

[0015] Determine the distance between each sample in the query data set and each of the category prototypes through the target prototype network;

[0016] Obtain the category prediction result output by the target prototype network, determine the error between the category prediction result and the training data according to a preset loss function, and adjust the parameters of the target generation model according to the error until the training is completed.

[0017] Furthermore, the loss function is obtained by weighted summation of a reconstruction loss function, a classification loss function and a relative distance loss function.

[0018] Furthermore, the weights corresponding to the reconstruction loss function, the classification loss function and the relative distance loss function are respectively determined by adaptive learning of the target prototype network.

[0019] According to another aspect of the present invention, there is provided a device for diagnosing faults of an ATM, comprising:

[0020] A data processing module for diagnosis, used for acquiring the data for diagnosis and processing the data for diagnosis;

[0021] The diagnosis result output module is used to use the processed data to be diagnosed as input data of the target prototype network, input the data to be diagnosed into the trained target prototype network, and obtain the diagnosis result output by the target prototype network.

[0022] Optionally, the to-be-diagnosed data processing module is further used for:

[0023] Performing continuous wavelet transform processing on the vibration signal of the banknote dispensing module to obtain a time-frequency diagram of the vibration signal;

[0024] The vibration signal time-frequency diagram is used as the processed data to be diagnosed.

[0025] Optionally, the target prototype network includes a feature extraction network, a distance measurement network, a fault diagnosis network and a decoder, wherein the feature extraction network acts as an encoder and forms a convolutional autoencoder with the decoder.

[0026] Optionally, the device further comprises a target prototype network training module, which is used to:

[0027] Acquire training data; wherein the training data includes a support data set and a query data set;

[0028] Performing type recognition on the support data set through the target prototype network to obtain at least one category prototype;

[0029] Determine the distance between each sample in the query data set and each of the category prototypes through the target prototype network;

[0030] Obtain the category prediction result output by the target prototype network, determine the error between the category prediction result and the training data according to a preset loss function, and adjust the parameters of the target generation model according to the error until the training is completed.

[0031] Optionally, the loss function is obtained by weighted summing a reconstruction loss function, a classification loss function and a relative distance loss function.

[0032] Optionally, the weights corresponding to the reconstruction loss function, the classification loss function and the relative distance loss function are respectively determined by adaptive learning of the target prototype network.

[0033] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0034] at least one processor; and

[0035] a memory communicatively connected to the at least one processor; wherein,

[0036] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the ATM fault diagnosis method described in any embodiment of the present invention.

[0037] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the ATM fault diagnosis method described in any embodiment of the present invention when executed.

[0038] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program / instructions, and the computer program / instructions, when executed by a processor, implement the steps of the ATM fault diagnosis method described in any embodiment of the present invention.

[0039] The ATM fault diagnosis method disclosed in the present invention first obtains data to be diagnosed and processes the data to be diagnosed; then the processed data to be diagnosed is used as input data of a target prototype network, and the data to be diagnosed is input into the trained target prototype network to obtain a diagnosis result output by the target prototype network. The ATM fault diagnosis method disclosed in the present invention can improve the fault diagnosis accuracy of the ATM, thereby reducing the possibility of business losses of bank outlets caused by ATM faults.

[0040] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 is a flow chart of a method for diagnosing ATM faults provided according to Embodiment 1 of the present invention;

[0043] Figure 2 is a schematic diagram of the structure of a target prototype network provided according to the first embodiment of the present invention;

[0044] Figure 3 is a structural schematic diagram of a teller machine fault diagnosis device provided according to Embodiment 2 of the present invention;

[0045] Figure 4 It is a structural schematic diagram of an electronic device for implementing the ATM fault diagnosis method of the third embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0047] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0048] Embodiment 1

[0049] Figure 1 This is a flowchart of a method for diagnosing a fault in an ATM provided in the first embodiment of the present invention. This embodiment is applicable to the case of diagnosing a fault in an ATM. The method can be performed by a device for diagnosing a fault in an ATM. The device for diagnosing a fault in an ATM can be implemented in the form of hardware and / or software. The device for diagnosing a fault in an ATM can be configured in an electronic device. Figure 1 As shown, the method includes:

[0050] S110, obtaining the data to be diagnosed, and processing the data to be diagnosed.

[0051] The data to be diagnosed are equipment-related data required for fault diagnosis of the ATM, such as vibration signal data of mechanical components such as rolling bearings and rollers contained in a fault module in the ATM.

[0052] For example, the cash dispensing module of an ATM contains many mechanical components such as rolling bearings and rollers. When customers go to a bank branch to handle deposit and withdrawal related services, the banknotes will pass through a conveyor belt composed of rolling bearings and other components. After the sensor detects that there is no adhesion, the ATM will dispense the banknotes. During the cash dispensing process, the vibration signal generated by the rolling bearing is collected and processed as the data to be diagnosed, so that the operating status of the cash dispensing module can be diagnosed in real time.

[0053] In this embodiment, after the data to be diagnosed is obtained, the data to be diagnosed may be screened, denoised, and processed to obtain processed data.

[0054] Optionally, the data to be diagnosed includes a vibration signal of a cash dispensing module, and the method for processing the data to be diagnosed may be: performing continuous wavelet transform on the vibration signal of the cash dispensing module to obtain a time-frequency diagram of the vibration signal; and using the time-frequency diagram of the vibration signal as the processed data to be diagnosed.

[0055] Specifically, the method for processing the vibration signal of the cash dispensing module may be to use continuous wavelet transform to process the vibration signal of the cash dispensing module into a vibration signal time-frequency diagram, and then use the obtained vibration signal time-frequency diagram as input data of the model for diagnosis.

[0056] S120, using the processed data to be diagnosed as input data of the target prototype network, inputting the data to be diagnosed into the trained target prototype network, and obtaining the diagnosis result output by the target prototype network.

[0057] Among them, the prototype network is a model for small sample classification, which can better represent each category by calculating the category prototype. The idea of ​​the prototype network is: for a classification task, an important goal is to find the category prototype of each category in the embedding space, and its prototype can better represent the corresponding category. In the prototype network, the samples of each category learn features through the embedding function. The embedding function uses learnable parameters to map the input sample to the embedding space, thereby obtaining the embedded features of the sample. For each category, its prototype can be obtained through multiple sample embedding features. Subsequently, the mean of the corresponding category embedding features is calculated as the category prototype. During the training process, the prototype network will learn a metric function, and the support set in the training set will find the prototype of the corresponding category through this metric function. A small number of samples in the query set can complete the classification task using the metric function and multiple category prototypes. During the testing process, the prototypes of each category are obtained through the support set of the test data, and the distance between the embedded features of the query set samples and the prototypes of each category is calculated. The prototype with the closest distance to the sample embedding feature is found, and its category is the category to which the query set samples belong.

[0058] Optionally, the target prototype network includes a feature extraction network, a distance measurement network, a fault diagnosis network and a decoder, wherein the feature extraction network serves as an encoder and a decoder to form a convolutional autoencoder.

[0059] Specifically, the target prototype network used in the embodiment of the present invention builds a network based on the prototype network in metric learning, and reduces the parameters that the network needs to learn through feature expression and similarity measurement, thereby reducing the impact of sample scarcity on the model. Figure 2It is a structural schematic diagram of a target prototype network provided by an embodiment of the present invention. As shown in the figure, in order to improve the feature extraction capability of the prototype network, the feature extraction network of the prototype network is used as an encoder, and multiple deconvolution modules are added as decoders to form a convolutional self-encoder. Among them, the encoder is composed of 4 convolutional modules, each of which includes a convolutional layer, a pooling layer, an activation layer, etc. Under the action of the 4 convolutional modules, the input data to be diagnosed can be compressed and reduced in dimension, thereby obtaining the embedded features of the sample. Corresponding to the encoder structure, the decoder is composed of 4 deconvolution modules to expand and map the features output by the encoder, and the encoder and the decoder constitute a convolutional self-encoder. In the decoder, data reconstruction is completed layer by layer, so that the feature dimensions of the corresponding layers of the encoder and the decoder are the same. It should be noted that the first three deconvolution modules of the decoder are composed of a deconvolution layer, a pooling layer, and a ReLU activation layer, and the fourth deconvolution module does not use a ReLU activation layer. This operation is to prevent the output of the decoder from not satisfying the normal distribution, resulting in an increase in the reconstruction error. After the decoder has completed the reconstruction of the data, the decoder output and the input of the entire model can be used to calculate the reconstruction error to measure the feature quality. At the same time, the output of the encoder is also used as the input of the distance measurement network and the fault diagnosis network. The prototype of each category is calculated through the distance measurement network, and the distance between each sample and the corresponding category prototype is calculated. Finally, the category corresponding to each sample, that is, the diagnosis result, is obtained through the fault diagnosis network. Optionally, the target prototype network can also include two fully connected layers, which are adaptive task weight calculation branches, used to calculate the weights of each item in the loss function of the target prototype network.

[0060] Optionally, the training method of the target prototype network may be: obtaining training data; wherein the training data includes a support data set and a query data set; performing type identification on the support data set through the target prototype network to obtain at least one category prototype; determining the distance between each sample in the query data set and each category prototype through the target prototype network; obtaining the category prediction result output by the target prototype network, determining the error between the category prediction result and the training data according to a preset loss function, and adjusting the parameters of the target generation model according to the error until the training is completed.

[0061] Specifically, in the training data, the K-way N-shot method can be used to randomly sample data to obtain the support set S and the query set Q. Among them, the support set S and the query set Q are used to calculate the prototype of each category and the distance between each sample and the corresponding category prototype, respectively. Subsequently, the support set S is used to calculate the prototype of each category, and the distance between each sample in the query set Q and all category prototypes is calculated to obtain the predicted category and calculate the loss function. In the process of model training, the back propagation algorithm is used to transfer the training error, and the stochastic gradient algorithm is used to update the network parameters to minimize the loss function. After the model training is completed, the verification data is passed into the model, and the model is trained according to the above steps to adjust the network parameters. After the target prototype network training is completed, the data to be diagnosed is passed into the trained target prototype network, and fault diagnosis and early warning can be implemented.

[0062] K-way N-shot is a sampling method in metric learning, which means randomly selecting K categories from the target data set, sampling N samples from each category, using K×N samples in the data set as the support set, and the remaining samples as the query set. In simple terms, the K-way N-shot problem is to complete the classification task of the query set samples with the reference of K×N samples.

[0063] Furthermore, the loss function is obtained by weighted summation of the reconstruction loss function, the classification loss function and the relative distance loss function.

[0064] Specifically, the present invention can use mean square error when calculating the reconstruction loss function. The reconstruction loss is composed of support set reconstruction loss and query set reconstruction loss. The reconstruction loss function is shown in the following formula:

[0065]

[0066] Among them, N s Represents the number of support set samples for each category, N Q represents the number of query set samples in each category, h and f represent the encoding and decoding calculation processes respectively, S i and Q k They represent the i-th sample in the support set and the k-th sample in the query set respectively.

[0067] After the convolutional autoencoder completes feature extraction, the output features of the encoder part are used as the input of the distance metric network. In the distance metric network, the prototype of each category is calculated using the support set samples, and then the distance between each sample and the corresponding category prototype is calculated using the query set. Finally, a category-based probability distribution is generated for the query set samples through the Softmax function, and the category with the highest probability is the category corresponding to the query set sample. The classification loss function of the prototype network can be obtained by calculating the negative logarithmic probability. The classification loss function is shown as follows:

[0068]

[0069] Among them, N c represents the number of categories, c i represents the category prototype of the ith category, d represents the distance metric function, and the Euclidean distance is used in the present invention.

[0070] Furthermore, when calculating the classification loss, the prototype network only considers the distance between the query set sample and the prototype of the same category, and does not consider the influence of the prototypes of different categories, which will reduce the accuracy of the category prototype to a certain extent and affect the accuracy of fault diagnosis. To solve this problem, the present invention designs a relative distance loss function, so as to incorporate the influence of prototypes of different categories into the model training process. Specifically, when measuring the distance, it is necessary to calculate the distance between the query set sample and the prototype of each category, including the intra-class distance and the inter-class distance. When the classification is correct, the intra-class distance is less than the inter-class distance, that is, the inter-class distance: the intra-class distance ≥ 1, otherwise the inter-class distance: the intra-class distance < 1, and the distance ratio can represent the correctness of the classification. Preferably, in order to make the intra-class distance as small as possible and the inter-class distance as large as possible, 1 can be used as the loss threshold, and the distance ratio less than 1 can be used as the loss. The smaller the distance ratio, the higher the degree of classification error, so 1 / distance ratio can be used as the loss, which is called relative distance loss. By introducing the relative distance loss, the prototype network can simultaneously consider the influence of the prototype of the category to which it belongs and the prototypes of different categories. The relative distance loss makes the distance between the sample and the prototype of the same category closer and the distance between the sample and the prototype of different categories farther, which can help the network generate more representative category prototypes, thereby improving the accuracy of the cash dispensing module fault diagnosis. The relative distance loss function is shown in the following formula:

[0071]

[0072] Among them, d ii represents the mean intra-class distance of the i-th category, d ij Represents the mean inter-class distance between the samples of the j-th category and the prototype of the i-th category.

[0073]

[0074]

[0075] Among them, x jk represents the kth sample of the i-th category.

[0076] The target prototype network used in this invention contains three tasks, namely, minimizing the reconstruction error of the convolutional autoencoder, minimizing the sample classification loss, and minimizing the relative distance loss. The three sub-loss functions constitute the loss function of this article, and the total loss function is shown as follows:

[0077] Loss = αLoss Recon +βLoss Cls +γLoss Dis

[0078] Among them, α, β, and γ represent task weights and are used to balance the losses of multiple tasks.

[0079] Furthermore, the weights corresponding to the reconstruction loss function, the classification loss function and the relative distance loss function are determined through adaptive learning of the target prototype network.

[0080] Specifically, in order to avoid improper manual selection of task weights, two fully connected layers (i.e. Figure 2 The adaptive task weight calculation shown in ) is used to train the task weight. The output dimension of the second fully connected layer is 3, and the output values ​​represent α, β, and γ respectively. The neural network continuously learns the parameters to achieve adaptive adjustment of the task weight.

[0081] The ATM fault diagnosis method provided by the embodiment of the present invention first obtains the data to be diagnosed and processes the data to be diagnosed; then the processed data to be diagnosed is used as the input data of the target prototype network, and the data to be diagnosed is input into the trained target prototype network to obtain the diagnosis result output by the target prototype network. The ATM fault diagnosis method disclosed by the present invention can improve the fault diagnosis accuracy of the ATM, thereby reducing the possibility of ATM faults causing business losses in bank branches.

[0082] Embodiment 2

[0083] Figure 3 A schematic diagram of the structure of a fault diagnosis device for an ATM provided in the second embodiment of the present invention is shown in FIG. Figure 3 As shown, the device includes: a diagnosis data processing module 310 and a diagnosis result output module 320.

[0084] The data to be diagnosed processing module 310 is used to obtain the data to be diagnosed and process the data to be diagnosed.

[0085] The diagnosis result output module 320 is used to use the processed data to be diagnosed as input data of the target prototype network, input the data to be diagnosed into the trained target prototype network, and obtain the diagnosis result output by the target prototype network.

[0086] Optionally, the to-be-diagnosed data processing module 310 is further used for:

[0087] The vibration signal of the banknote dispensing module is processed by continuous wavelet transform to obtain a time-frequency diagram of the vibration signal; the time-frequency diagram of the vibration signal is used as the processed data to be diagnosed.

[0088] Optionally, the target prototype network includes a feature extraction network, a distance measurement network, a fault diagnosis network and a decoder, wherein the feature extraction network serves as an encoder and a decoder to form a convolutional autoencoder.

[0089] Optionally, the device further includes a target prototype network training module 330, which is used to:

[0090] Acquire training data; wherein the training data includes a support data set and a query data set; perform type recognition on the support data set through a target prototype network to obtain at least one category prototype; determine the distance between each sample in the query data set and each category prototype through the target prototype network; obtain the category prediction result output by the target prototype network, determine the error between the category prediction result and the training data according to a preset loss function, and adjust the parameters of the target generation model according to the error until the training is completed.

[0091] Optionally, the loss function is obtained by weighted summing the reconstruction loss function, the classification loss function and the relative distance loss function.

[0092] Optionally, the weights corresponding to the reconstruction loss function, the classification loss function, and the relative distance loss function are respectively determined by adaptive learning of the target prototype network.

[0093] The ATM fault diagnosis device provided in the embodiment of the present invention can execute the ATM fault diagnosis method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0094] Embodiment 3

[0095] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0096] like Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0097] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0098] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a teller machine fault diagnosis method.

[0099] In some embodiments, the ATM fault diagnosis method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the ATM fault diagnosis described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the ATM fault diagnosis method in any other appropriate manner (e.g., by means of firmware).

[0100] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0101] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0102] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0103] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0104] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0105] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0106] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0107] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for diagnosing a fault of an ATM, characterized in that: include: Acquiring data to be diagnosed, and processing the data to be diagnosed; The processed data to be diagnosed is used as input data of the target prototype network, and the data to be diagnosed is input into the trained target prototype network to obtain the diagnosis result output by the target prototype network.

2. The method according to claim 1, characterized in that The data to be diagnosed includes a vibration signal of a banknote dispensing module, and processing the data to be diagnosed includes: Performing continuous wavelet transform processing on the vibration signal of the banknote dispensing module to obtain a time-frequency diagram of the vibration signal; The vibration signal time-frequency diagram is used as the processed data to be diagnosed.

3. The method according to claim 1, characterized in that The target prototype network includes a feature extraction network, a distance measurement network, a fault diagnosis network and a decoder, wherein the feature extraction network serves as an encoder and forms a convolutional autoencoder with the decoder.

4. The method according to claim 3, characterized in that The training method of the target prototype network includes: Acquire training data; wherein the training data includes a support data set and a query data set; Performing type recognition on the support data set through the target prototype network to obtain at least one category prototype; Determine the distance between each sample in the query data set and each of the category prototypes through the target prototype network; Obtain the category prediction result output by the target prototype network, determine the error between the category prediction result and the training data according to a preset loss function, and adjust the parameters of the target generation model according to the error until the training is completed.

5. The method according to claim 4, characterized in that The loss function is obtained by weighted summation of a reconstruction loss function, a classification loss function and a relative distance loss function.

6. The method according to claim 5, characterized in that The weights corresponding to the reconstruction loss function, the classification loss function and the relative distance loss function are respectively determined by adaptive learning of the target prototype network.

7. A teller machine fault diagnosis device, characterized in that: include: A data processing module for diagnosis, used for acquiring the data for diagnosis and processing the data for diagnosis; The diagnosis result output module is used to use the processed data to be diagnosed as input data of the target prototype network, input the data to be diagnosed into the trained target prototype network, and obtain the diagnosis result output by the target prototype network.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the ATM fault diagnosis method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the ATM fault diagnosis method according to any one of claims 1 to 6 when executed.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the ATM fault diagnosis method according to any one of claims 1 to 6 are implemented.

Images