6 results about "Sample average" patented technology

The invention relates to the technical field of financial science and technology, in particular to a model performance prediction method and a performance prediction model construction method and device. The method comprises the following steps: determining influence factors of a to-be-predicted model; inputting the influence factors into a performance prediction model to obtain a resource consumption key prediction value and an average time consumption key prediction value required by calling a to-be-predicted model, the performance prediction model constructing a training sample set by a sample resource consumption key prediction value of a sample model, a sample average time consumption key prediction value and a plurality of influence factors, training a plurality of initial prediction models; and according to the resource consumption key prediction value and the average time consumption key prediction value, determining a resource consumption overall prediction value and an average time consumption overall prediction value of the to-be-predicted model. According to the scheme, before the model is developed, the resource overhead and performance after the model is online can be predicted, the labor burden is reduced, reworking is avoided, and the model delivery efficiency is ensured.

The invention discloses a multi-period random optimal power flow problem solving method, model construction is performed on the optimal power flow problem of a system based on two-stage stochastic programming, and a sample average approximation method is further adopted in the embodiment to convert the optimal power flow problem from a stochastic optimization problem into a deterministic optimization problem. The neural network is introduced for solving the sub-problems of the two stages to complete solving of decision variables, the trained neural network approximates a mapping relation of a complex optimization problem, the problem of high calculation cost caused by large sample size and multiple sub-problems in traditional two-stage stochastic programming can be remarkably relieved, and compared with a traditional solving method, the method is more efficient, and the calculation cost is reduced. And considerable accuracy is maintained.

The invention relates to the technical field of load control, in particular to a heavy load inquiry control method and system for a ship power management system, and the method comprises the steps: introducing a Monte Carlo algorithm to construct a state transition structure, setting a state behavior sample in a capacity interval, and setting a transition path under the combination of a sampling action and capacity; capacity consumption responses under all actions are accumulated, an action benefit matrix with quantifiable weights is generated through a sample average calculation mode, a data-driven load response model is formed, the clearness of scheduling action evaluation is improved, after accumulation traversal is conducted on a full-state space, the action return change amplitude is recognized, and the scheduling action evaluation accuracy is improved. And on the basis of state convergence, a greedy algorithm is called to read a corresponding maximum benefit action, an action scheduling sequence is constructed in sequence according to state numbers, the symmetry and the integrating degree of power configuration are improved, the action mismatching rate and the distribution deviation rate are reduced, and the power scheduling convergence efficiency is improved.

The utility model provides a kind of continuous granulating device for sample average weight determination, including multi-bin type box and install on the continuous material pushing mechanism of multi-bin type box, and the continuous material pushing mechanism of multi-bin type box includes storage bin and the discharge bin being set to the bottom of storage bin, the mould plate of the bottom of storage bin is equipped with blanking hole, and the left end of discharge bin is set as discharge port;The one end of continuous material pushing mechanism is located in the right end of discharge bin, and is located in the side position of blanking hole, to be measured sample from blanking hole under the reciprocating push of another end falls from the outlet of discharge bin continuously pushes out.The utility model can continuously push out the sample to be measured from the blanking hole of the bottom of storage bin by continuous material pushing mechanism, can satisfy the sampling demand of continuously determining the weight of multiple (granule);And the continuous granulating device structure design is simple, novel, convenient to use, for sampling determination, can effectively improve operating efficiency.

The invention discloses a deep learning method for an outage probability optimization problem, and belongs to the field of deep learning. For a downlink transmission scene of the MISO system, modeling is carried out by taking minimization of the total outage probability of all users as an optimization target, an indicative function is introduced, the conditional outage probability of each user in modeling is converted into a conditional expectation form, a sample average method is adopted to approximate a conditional expectation value, and a smooth function is adopted to approximate the indicative function; then, constructing a graph neural network for learning a precoding strategy by using a smoothing function and unsupervised learning, designing an adaptive smoothing function method, modeling parameters of the smoothing function, and solving by using binary search; and then, alternately updating the graph neural network and smooth function parameters until the training achieves convergence. And finally, inputting an estimation channel obtained in real time into the optimized graph neural network, outputting a learned precoding strategy, and ensuring that the total outage probability of all users is minimized. The method is low in calculation complexity.

Provided are a bolted truss steel bridge member detection method, device and readable storage medium, covering full-dimensional detection of "port, node and overall structure"; adapting to the core features of the bolted structure "hole connection"; constructing a standardized detection system of "coding, extraction and detection"; designing attribute-embedded coding rules to directly analyze the key attributes of the member through coding; establishing a fixed process of "targeted extraction, multi-dimensional detection and error determination"; designing a feature point extraction method guided by "assembly sequence"; and designing a "circle fitting calculation" and "linear fitting combined with sampling average" quantitative algorithm to convert the structure stability index into a directly comparable numerical result. The method comprises S1, basic framework building, S2, calculation item point extraction and S3, truss piece parameter detection. The application realizes comprehensive, efficient, standardized and accurate detection of the detection process by automatically matching the member features through coding analysis, and solves the technical problems of manual intervention errors and low work efficiency.